If you find our site valuable and helpful, consider supporting us or sponsoring one of our curations.
Source/Link | Description |
|---|---|
Information Week - The AI infrastructure boom is coming for enterprise budgets | The AI boom has largely been discussed as a story of productivity and competitive advantage. But the economics underpinning it all are becoming harder to ignore. Training frontier models, scaling inference workloads, supporting AI agents, and maintaining increasingly compute-intensive enterprise features requires enormous infrastructure investment, from GPUs and networking equipment to data centers and energy consumption. |
InfoWorld - 12 model-level deep cuts to slash AI training costs | Optimizing artificial intelligence pipelines requires moving beyond surface-level hardware adjustments to fundamentally alter how models process data. While engineers often implement basic toggle-away efficiencies inside the training loop, achieving permanent cost reductions requires architectural changes directly inside the neural network. As I have previously argued, the science is solved, but the engineering is broken; true FinOps maturity demands deep, model-level interventions. The following 12 architectural cuts will drastically lower the unit economics of your AI pipeline |
Information Week - AI on trial: The Workday case that CIOs can't ignore | Some 14,000 people have recently opted in to a case that is effectively putting AI hiring systems on trial. The participants are all at least 40 years old and claim they were unfairly denied jobs after being screened by Workday's recruiting systems that score, sort and rank applicants. |
VentureBeat - Intent-based chaos testing is designed for when AI behaves confidently — and wrongly | What makes this scenario particularly uncomfortable is that the failure was not in the model. The model behaved exactly as trained. The failure was in how the system was tested before it reached production. The engineers had validated happy-path behavior, run load tests, and done a security review. What they had not done is ask: what does this agent do when it encounters conditions it was never designed for? |
SiliconANGLE - Nvidia, AI factories and the transition to accelerated computing | Nvidia is creating a new platform by becoming the default substrate for enterprise computing, and that platform pulls everything else into its ecosystem. In this cycle, the unit driving growth isn’t a PC or a server, it’s the artificial intelligence factory, which is a rack-scale system that turns power, data, compute and software into intelligence through tokens, reasoning and automated workflows. |
Information Week - As AI makes projects harder to track, will CIOs need new controls? | The issues arise immediately. Unlike traditional projects, AI initiatives don't necessarily begin in IT, said Jen Clark, director of AI advisory services at Eisner Advisory Group. "They start within the business whenever someone finds or builds a tool that solves a problem," she said. This leaves CIOs without clear visibility from Day 1. And unfortunately, the flow of scaled rollout hasn't changed to match the speed, coverage and capability of these tools. |
VentureBeat - AI tool poisoning exposes a major flaw in enterprise agent security | If you’re using agents that choose tools from centralized registries, add endpoint allowlisting as a bare minimum today. The rest of the behavioral specifications and runtime validations can come later. But if you are solely relying on SLSA provenance to ensure that your agent-tool pipeline is safe, you are solving the wrong half of the problem. |
SiliconANGLE - Red Hat targets enterprise deployment with new version of its AI platform | The Red Hat AI strategy is divided into four key pillars, said Joe Fernandes, vice president and general manager of Red Hat AI. “First, helping customers deliver fast, flexible and efficient inference, serving models in their environment,” he said in a pre-event briefing. “Second, connecting their enterprise data to those models and agents. Third, helping them accelerate the deployment and management of agents across a hybrid cloud environment. Fourth, bringing that all together on our integrated AI platform, enabling them to run any model in any agent across any hardware and cloud environment.” |
Too many enterprises still treat the debate of cloud versus on-premises as a purely technical decision. It is not. It is a business decision, an operating model decision, a governance decision, and, increasingly, a supply chain decision. If the price of memory rises as hyperscalers vacuum up supply to support AI expansion, the cloud may appear cheaper in the short term. But cheaper under those conditions does not mean better. It means the baseline has shifted. | |
ZDNet - Give your 'human-level agents' a proper head start with these 3 best practices | "They have this challenge, which is they want to offer stronger and stronger feedback and advice and guidance and insight to their app users," Wiley explained. "But they need to be unbelievably careful because this is very sensitive data, so the last thing they would want is an app user to get a response that includes some other app user's information in it." |
SiliconANGLE - Nvidia’s MRC: When ‘just Ethernet’ isn’t enough for gigascale AI | On the surface, MRC is a new remote direct memory access or RDMA transport protocol, now open-sourced via the Open Compute Project. In reality, it’s a production-proven way to keep tens or hundreds of thousands of graphics processing units fed and synchronized by using a single RDMA connection to stripe traffic across multiple paths and dynamically steer around congestion and failures. OpenAI has already used MRC on Spectrum-X to train recent frontier large language models powering ChatGPT and Codex, and Microsoft is deploying it in some of its largest AI factories built on GB200 systems. The important point is that MRC isn’t a lab experiment but a set of algorithms that has already earned its place in some of the most demanding AI environments on the planet. |
InfoWorld - Building AI apps and agents with Microsoft Foundry | Microsoft Foundry helps application developers to build and deploy agents, which may use models and tools. It also helps machine learning (ML) engineers and data scientists to fine-tune models, run evaluations, and manage model deployments. Finally, it helps IT administrators and platform engineers to govern AI resources, enforce policies, and manage access across teams. It isn’t quite a floor wax and a dessert topping, but it does try to serve three distinct audiences. |
The scanner never looked at the .test.ts file sitting one directory over. It didn’t need to. Test files aren’t part of the agent execution surface, so no publicly documented scanner inspects them (as of publication of this post). The file runs anyway. Not through the agent but through the test runner, with full access to the filesystem, environment variables, and SSH keys. | |
Founded in 2023, OpsMill develops software that gives enterprise infrastructure and network teams a unified, structured view of their IT environments so engineers and AI agents can automate operations safely at large scale. Its flagship product, Infrahub, uses a graph database to map connections between hundreds of thousands of infrastructure elements across physical, virtual and cloud-based environments, serving as what the company calls a trusted source of truth for any IT environment. | |
This is often one of the biggest mindsets shifts for frontend engineers. We often think about failure as a total outage where the whole site is down. In practice, that is not what most users experience. More often, the interface is partially degraded: A dashboard loads but one panel is empty, a form saves but the confirmation never arrives, or a file upload stalls while the rest of the page still appears normal. | |
VentureBeat - Microsoft takes Agent 365 out of preview as shadow AI becomes an enterprise threat | The product, first announced at Microsoft's Ignite conference in November, positions itself as a unified control plane that lets enterprise IT and security teams observe, govern, and secure AI agents wherever they run: inside Microsoft's own ecosystem, on third-party cloud platforms like AWS Bedrock and Google Cloud, on employee endpoints, and increasingly across a sprawling ecosystem of SaaS agents built by partner software companies. |
SiliconANGLE - Collibra’s new AI Command Center promises to combat agentic hallucinations | As Collibra Chief Executive Felix van Van de Maele explains, almost every enterprise that’s trying to use AI agents has to pay a kind of “hallucination tax,” which refers to the hidden costs of manual oversight, correcting agents’ mistakes and the general risk that things might go wrong. “The AI Command Center eliminates that tax,” he promised. “It gives organizations real-time visibility, continuous control and the confidence to run AI at the speed it actually moves.” |
Agent wranglers are required to bring management sensibilities to this growing space. So, can AI agent sprawl be tamed? Some vendors are giving it a try, leading to a new technology category, agent management systems, that are tasked with managing networks of AI agents. | |
CLI-Anything generates SKILL.md files, the same instruction-layer artifacts that Snyk’s ToxicSkills research found laced with 76 confirmed malicious payloads across ClawHub and skills.sh in February 2026. A poisoned skill definition does not trigger a CVE and never appears in a software bill of materials (SBOM). No mainstream security scanner has a detection category for malicious instructions embedded in agent skill definitions, because the category simply did not exist eighteen months ago. | |
Computerworld - Edge browser leaves passwords exposed in plain text, says researcher | Microsoft has been nonchalant about the discovery. It said, “Design choices in this area involve balancing performance, usability, and security, and we continue to review it against evolving threats. Browsers access password data in memory to help users sign in quickly and securely — this is an expected feature of the application.” |
The world runs on open-source software. We all know that. But did you know that companies download over 10 trillion (that's trillion with a T) open-source code files every year? According to software security provider Sonatype, they do --and the file repository sites that supply that code are burning out from the demand. | |
InfoWorld - Diskless databases: What happens when storage isn’t the bottleneck | Diskless architectures sidestep traditional constraints by separating compute from storage and removing local persistence from the critical path. Data is ingested and indexed in memory for immediate availability, while object storage provides the durable, elastic foundation underneath. The result is a database that accelerates both ingestion and retrieval without sacrificing persistence. |
VentureBeat - Microsoft takes Agent 365 out of preview as shadow AI becomes an enterprise threat | But the most striking element of the launch isn't the general availability milestone itself. It's Microsoft's aggressive push into discovering and managing local AI agents — the coding assistants, personal productivity tools, and autonomous workflows that employees are installing on their own devices, often without IT's knowledge or blessing. Microsoft calls this phenomenon "shadow AI," and it is an entirely new category of enterprise security risk that most organizations are only beginning to grapple with. |
SiliconANGLE - WSO2 launches Agent Manager to help enterprises tame AI agent sprawl | WSO2 Agent Manager comes as organizations are accelerating agentic AI adoption to avoid being left behind, driven by the promise of nonlinear productivity gains that agentic systems can unlock. WSO2, however, argues that operational maturity is lagging behind, with teams forced to choose between moving fast with limited visibility and control, introducing significant unmanaged risk, or slowing progress to build operational frameworks for each runtime and environment. |
ZDNet - What you'll pay for AI agents will be wildly variable and unpredictable | You might expect agents to cost more in tokens, but the study reveals more alarming facts. Two different models can have wildly different token costs for the same task. And the same model can have different costs each time that it works on the same problem, using as many as twice the number of tokens on one occasion compared to another. |
SiliconANGLE - IBM charts AI operating model to move enterprises beyond experimentation | The announcements span agent orchestration, real-time data integration, hybrid cloud operations and digital sovereignty, reflecting what executives described as a shift away from isolated AI deployments toward systemic integration across the enterprise. |
Computerworld - Microsoft, Google push AI agent governance into enterprise IT mainstream | “By placing agent controls alongside identity, access, data, and workload management, vendors are positioning AI governance as an operational discipline owned jointly by IT and security,” said Biswajeet Mahapatra, principal analyst at Forrester. “For CIOs, this means AI agents now need to be managed like any other digital workforce, with lifecycle oversight, cost visibility, and integration into service management.” |
ZDNet - Building an agentic AI strategy that pays off - without risking business failure | The first choice will barely move the needle, but will help the AI initiative pay for itself. The second choice could blow the doors off your numbers and make you a legend in your board's eyes. It could also get you fired. |
Computerworld - Relying on LLMs is nearly impossible when AI vendors keep changing things | Over the years, enterprise IT execs have gotten frighteningly comfortable having little control or visibility over mission-critical apps, from SaaS to cloud and even cybersecurity. But generative AI (genAI) and agentic systems are taking that problem to a new extreme, with vendors able to dumb down a system IT is paying billions for without so much as a postcard. |
ZDNet - The future of IT service delivery is built on AI and automation | For efficient scaling, layering on IT complexity is no longer sustainable. In one survey, more than half of managed service providers (MSPs) say they experience vendor sprawl, and the damage from the resulting operational drag goes far beyond lost efficiency and margins. It also breaks visibility, as techs spend crucial time between isolated platforms and face fatal blind spots in security posture. |
Computerworld - AI agents can bypass guardrails and put credentials at risk, Okta study finds | It’s no secret that AI agents have huge potential, balanced by equally big risks. What’s becoming apparent, however, is how quickly agentic systems can veer wildly off course and start exposing critical information under real-world conditions. |
SiliconANGLE - AI exposes attacks traditional detection methods can’t see | Side-channel attacks don’t provide that. Neither do many modern intrusions. An attacker operating through encrypted channels, legitimate tools or AI-assisted workflows can move through an environment without ever triggering a condition that a rule can evaluate. The activity is valid at every individual step. The pattern becomes visible only when you look at how those steps connect over time. |
VentureBeat - 200,000 MCP servers expose a command execution flaw that Anthropic calls a feature | Anthropic created the Model Context Protocol as the open standard for AI agent-to-tool communication. OpenAI adopted it in March 2025. Google DeepMind followed. Anthropic donated MCP to the Linux Foundation in December 2025. Downloads crossed 150 million. Then four researchers at OX Security found an architectural problem that affects all of them |
We do not believe the industry should over-rotate on Mythos as a standalone event. OpenAI Group PBC, Google LLC and other leading model providers have the technical capacity, research depth and, in some cases, greater compute capacity, to deliver similar capabilities. Tasks that once required scarce expert labor – code review, dependency analysis, configuration assessment and exploit-path discovery – are becoming cheaper, faster, automated and more scalable. | |
VentureBeat - Salesforce launches Agentforce Operations to fix the workflows breaking enterprise AI | Enterprise AI teams are hitting a wall — not because their models can't reason, but because the workflows underneath them were never built for agents. Tasks fail, handoffs break, and the problem compounds as organizations push agents deeper into back-office systems. A new architectural layer is emerging to address it: workflow execution control planes that impose deterministic structure on processes agents are expected to run. |
It doesn’t just compute; it generates reasoning, language, recommendations and even makes decisions. No one consciously gives up judgment; it simply becomes easier not to exercise it. After all, judgment demands reflection and diligence, AI helps us bypass this cognitive load in seconds. | |
Amazon Web Services on Tuesday launched one of the most consequential enterprise AI plays in the company's 20-year history, simultaneously bringing OpenAI's most powerful models to its Bedrock platform, unveiling a new agentic developer framework, releasing a desktop AI productivity tool called Amazon Quick, and expanding its Amazon Connect service from a single contact-center product into a family of four agentic AI solutions targeting supply chains, hiring, healthcare, and customer experience. | |
Information Week - Anthropic's Mythos forces a rethink of vulnerability management | Mythos is being used by Anthropic and Project Glasswing to identify and exploit zero-day vulnerabilities in open source codebases. Anthropic's own testing of Mythos uncovered that the AI is "capable of identifying and then exploiting zero-day vulnerabilities in every major operating system and every major web browser when directed by a user to do so." The Mythos tests even identified some vulnerabilities that are over 20 years old. In addition, less than 1% of potential vulnerabilities uncovered by Mythos have been fully patched by their maintainers, according to Gartner. |
Enterprise intent to adopt hybrid retrieval tripled from 10.3% to 33.3% in a single quarter — even as 22% of qualified enterprise respondents reported having no production RAG systems at all. For data engineers and enterprise architects building agentic AI infrastructure, the data reveals a market in active transition: the RAG architecture most enterprises built to scale is not the one they expect to run by year-end. | |
SiliconANGLE - DigiCert debuts AI Trust framework to secure agents, models and content | To address the challenge, DigiCert is introducing a unified trust layer that spans AI agents, models and content that embeds cryptographic verification across the AI lifecycle. With the release, organizations can enforce identity-based governance for autonomous systems, validate model integrity and establish content provenance all within a single, cohesive framework. |
One of the more dangerous assumptions in the current AI market is that broad adoption means meaningful adoption. It does not. Much of what enterprises call AI transformation is, in fact, AI experimentation focused at the edge of the business, in systems and workflows that support employees but are not central to how the enterprise actually operates. These include calendaring, scheduling, meeting summaries, employee communications, customer messaging, document generation, internal assistants, and similar productivity-oriented use cases. | |
VentureBeat - FOMO is why enterprises pay for GPUs they don't use — and why prices keep climbing | That pressure — repeated across thousands of enterprises over the past two years — is the reason most companies are now running their GPU fleets at roughly 5% utilization, according to Cast AI's 2026 State of Kubernetes Optimization Report, which measured actual production clusters rather than surveying them. It's also the reason nobody releases the idle capacity. Cast AI co-founder and President Laurent Gil has been tracking the dynamic for two years. “Many of the neoclouds are not cloud,” he told VentureBeat. “They are neo-real estate.” |
SiliconANGLE - Panzura opens its global filesystem to Microsoft Copilot users | The Nexus platform integrates with Panzura’s CloudFS hybrid cloud file system to expose data through Copilot’s conversational interface. The goal is to allow users to query enterprise knowledge using natural language without changing existing workflows or data architectures. |
InfoWorld - Critical GitHub RCE bug exposed millions of repositories | Uncovered by Wiz researchers, the now-patched bug exploited how GitHub handles server-side “git push” operations. By crafting malicious input within a standard Git push, an authenticated user could execute arbitrary commands via GitHub’s backend Git processing pipeline. |
Computerworld - SAS makes AI governance the centerpiece of its agent strategy | Its capabilities include: General Q&A across core Viya applications; production of documented and explainable AI-generated code; model pipeline guidance including recommendations and next steps; conversational dashboarding; and visual investigation with AI-assisted search and alert narratives. Copilot capabilities will eventually extend to data management, model management, and AI infrastructure, according to SAS. |
SiliconANGLE - Salesforce introduces Agentforce Operations to automate outdated back-office tasks | The company said the entire engine operates on “radical transparency,” meaning every action the AI takes is recorded to an audit trail, allowing IT teams to guide the agents, find out where things went wrong and quickly get back on track. |
Fullstack HR Substack - Stop Calling It a Pilot - by Johannes Sundlo - FullStack HR | People walk into workshops with problems they have been waiting on IT to solve for eighteen months. Eighteen months. By the end of a session some of them have a working tool that solves the problem. Not a perfect tool, but a working one. Enough to use Monday morning |
SiliconANGLE - Lookout launches mobile-native tool to expose shadow AI on enterprise devices | “AI adoption is accelerating faster than most organizations can see or control, especially on mobile, where AI activity often operates outside traditional corporate boundaries and remains largely invisible,” said Chief Executive Jim Dolce. “With the launch of Lookout AI Visibility & Governance, we’re closing that gap, giving organizations the ability to see, understand and govern AI usage at the mobile layer, bringing mobile AI activity out of the shadows into full visibility and control.” |
Computerworld - EU lawmakers fail to agree on watered-down AI Act, talks pushed to May | CIOs should treat August 2 as a hard deadline regardless of what happens in May, Shah said. “I believe CIOs are in a tough spot right now. They should be prepared, irrespective of the regulatory limbo, and treat this summer as a hard deadline. If it gets delayed, then it’s a bonus and if not, then it would be a regulatory risk.” |
SiliconANGLE - Aviatrix launches AI agent containment platform for cloud workloads | AI agents introduce a different kind of security problem because they do not need to be “broken into” in the traditional sense to become dangerous. An agent can be manipulated through prompt injection, where malicious instructions are hidden in content the agent reads, or through model poisoning, where the data or tools it depends on are corrupted. |
The Nuanced Perspective Substack - The AI Agent Stack in 2026 - by Aishwarya Naresh Reganti | The stack looks incredibly different today. In fact, the term stack itself is starting to feel like a narrow term for what the agent layer has become. It implies one direction of dependency: build on top of what is below. The shape that has been more useful lately looks more like an operating system than a stack. Picture it from the top. A surface where humans meet the agent. Beneath that, a core where the agent’s loop actually runs. Two supporting layers next to the core for knowledge and memory. Models at the bottom. Running down both sides of all of it, two rails that govern what the agent is allowed to do and how its work gets reviewed. |
The paper, "Training for Compositional Sensitivity Reduces Dense Retrieval Generalization," tested what happens when teams train embedding models for compositional sensitivity. That is the ability to catch sentences that look nearly identical but mean something different — "the dog bit the man" versus "the man bit the dog," or a negation flip that reverses a statement's meaning entirely. That training consistently broke dense retrieval generalization, how well a model retrieves correctly across broad topics and domains it wasn't specifically trained on. Performance dropped by 8 to 9 percent on smaller models and by 40 percent on a current mid-size embedding model teams are actively using in production. | |
DataCenter Knowledge - The Breaking Points: Networking Strains Under AI’s Scale Demands | “In the data center, people are finding it difficult to integrate the new networks they build out for AI with legacy infrastructure,” said Shamus McGillicuddy, vice president of research at Enterprise Management Associates. “This might include protocol mismatches – network borders where RoCE collides with TCP/IP.” |
Aranya’s ClusterdOS fills the gap between Kubernetes, distributed systems and AI infrastructure with an open-source, distributed operating system that turns raw compute into batteries-included, ready-for-production AI supercomputers. ClusterdOS handles the full cluster lifecycle, including bootstrapping, maintaining and upgrading with minimal effort and provides a straightforward framework for adding and versioning distributed cloud-native applications, configurable through simple high-level feature flags. | |
VentureBeat - Context decay, orchestration drift, and the rise of silent failures in AI systems | The most expensive AI failure I have seen in enterprise deployments did not produce an error. No alert fired. No dashboard turned red. The system was fully operational, it was just consistently, confidently wrong. That is the reliability gap. And it is the problem most enterprise AI programs are not built to catch. |
ZDNet - 77% of IT managers say their AI agents are out of control - 5 ways to rein in yours | That's the conclusion of a just-released survey by Rubrik ZeroLabs, which finds that fewer than one in four IT managers (23%) say they have "complete" control over the agents within their organizations. To make matters worse, these agents aren't necessarily delivering the productivity sought. A majority, 81%, report that the agents under their purview require more time in manual auditing and monitoring than they were intended to save via workflow improvements. Security is also less than stellar, the survey adds. |
Both CVEs sit on the CISA Known Exploited Vulnerabilities catalog. Neither score flagged the kill chain. The triage logic that consumed those scores treated each CVE as an isolated event, and so did the SLA dashboards and the board reports those dashboards feed | |
With today’s updates, Appian says, it’s providing its AI agents with more structure, context and guardrails, making them even smarter, more reliable and increasingly effective. The biggest change is the adoption of MCP, which enables its agents to securely interface with external enterprise systems. At the same time, the integration means third-party agents will be able to access Appian’s platform, including tools such as its Data Fabric offering, which provides read-write access to enterprise process data. | |
The AI Memo Substack - The AI Cost Most Companies Miss - by Andreas Welsch | On paper, open-source frameworks or lower-cost standalone tools like LibreChat can look much cheaper than Copilot or enterprise ChatGPT/Claude licenses. But many companies overlook the hidden costs that arise when employees end up doing the integration manually. That became immediately apparent in the interactions with leaders. |
SiliconANGLE - Startup Lovelace targets contextual AI engine at mission-critical use cases | At the core of the company’s approach is a “context engine” builder called Elemental that’s designed to sit between AI agents and underlying data systems. Elemental creates secure, enterprise-specific context engines that create structured knowledge graphs from fragmented data that AI agents can navigate and query to return research-quality analysis with citations. |
InformationWeek - CIOs caught in the middle as AI startups disrupt vertical Saas | As AI-native startups step into this scene, it raises questions about the continued dominance of vertical SaaS tools. "These startups aren't disrupting vertical SaaS at the system of record level, at least not yet," said Ayush Raj Jha, senior software engineer at Oracle. Instead of presenting a direct, one-for-one challenge to the incumbent technology, he said AI startups make the workflow layer above those systems irrelevant. "That's actually the more dangerous threat to SaaS." |
Computerworld - Microsoft, OpenAI change contract terms — again | Microsoft and OpenAI on Monday again revised their agreement, softening their exclusivity and revenue-sharing conditions in the process. These changes underscore how critical it is for enterprises to work with as many AI vendors as practical, given the leapfrogging performance stats as well as the constantly shifting alliances. |
ZDNet - GitHub Copilot shifts to usage-based pricing June 1 - why that's no surprise | This is a radical change from its current premium request unit (PRU) system. Going forward, users will consume monthly allotments of GitHub AI Credits based on token consumption, including input, output, and cached tokens at published API rates. In other words, GitHub is moving to a token-based pricing model. |
Computerworld - Xiaomi releases MIT‑licensed MiMo models for long‑running AI agents | By using the MIT License, Xiaomi said it is allowing commercial deployment, continued training, and fine-tuning without additional authorization. Tulika Sheel, senior vice president at Kadence International, said the MIT License can make it attractive. “It allows enterprises to freely modify, deploy, and commercialize the model without restrictions, which is rare in today’s AI landscape,” Sheel said. |
Fabrix’s platform runs an identity knowledge graph that analyzes access activity, organizational context and intent, paired with AI agents that handle authorization decisions, just-in-time access requests and full identity lifecycle management. The approach is intended to replace the manual, rules-based reviews that have traditionally governed enterprise access control with continuous, context-aware decisioning. | |
The goal shouldn’t be to ban AI or replace employees outright, but to use AI to cultivate a powerful knowledge ecosystem that captures knowledge, facilitates its movement, and creates new understanding. (Think Slack channels, wikis, tribal knowledge, onboarding docs, expert networks, and AI layers on top.) | |
SiliconANGLE - Google’s AI agent platform takes pole position but work remains | We think measurable business value will ride on top of this infrastructure and that is where the real battle lines will be drawn. Specifically, we see frontier model vendors, of which Google is one, rapidly building out capabilities that will become fundamental to the future of software – which we predict will be the biggest transformation in the history of the software industry. |
ZDNet - Nearly half of cybersecurity pros want to quit - here's why | Unsurprisingly, the survey found that security specialists have had enough. People working in cybersecurity are the third-most unhappy IT professionals globally (23%), just behind those working in quality assurance/testing (24%) and infrastructure/support (25%). |
SiliconANGLE - From GPUs to AI factories: Inside the Nvidia-Google Cloud superstack | The second point is that there should be plenty of capacity for AI factories. The technology footprint that underpins Google’s AI Hypercomputer concept — multitenant, massively scaled clusters where training, fine-tuning and inference share the same fabric — makes it realistic for enterprises to spin up large language model and agent workloads that run across tens of thousands of Nvidia GPUs without bespoke infrastructure engineering |
I would imagine that if you were in the business of analyzing data and providing dashboard-level insights into that data, then you would be very worried indeed about what AI is going to do to your value proposition. Much of the SaaS industry is in the business of analyzing existing data, and that is exactly what AI agents can do well. When a simple question can get straight to the heart of what a pricey dashboard provides, then companies have to question the value of paying for that kind of service. | |
SiliconANGLE - Software artifact management startup Cloudsmith raises $72M | Another use case that Cloudsmith supports is storing software containers. A container can comprise upwards of dozens of individual artifacts, each of which represents a potential cybersecurity risk. Cloudsmith tackles that complexity by automatically generating a software bill or materials, or SBOM, for each container. A SBOM is a file that lists a workload’s components. |
In this new world, the philosophy, ethics, and morals of open source are more relevant than ever. However, the focus of open source needs to evolve past raw code: Specification files (spec files) and governance documents (constitutions) are becoming as important as the source itself. The challenge is not to choose between open source and AI, but to recognize that open source is now a community-based control and scope mechanism for open technologies. | |
SiliconANGLE - Portal26 launches Agentic Token Controls to cap runaway AI agent spend | Portal26 argues that multistep autonomous agents built on large language models can unintentionally enter recursive loops, over-query systems or expand tasks beyond their original scope, leading to exponential token usage and surprise bills. The company is pitching the new module as the first dedicated tool to manage the risk at enterprise scale. |
InfoWorld - Microsoft taps Anthropic’s Mythos to strengthen secure software development | The announcement comes as Anthropic’s Mythos heightens concerns that advanced AI models could dramatically shrink the time between finding a software flaw and exploiting it. Analysts say Mythos marks a notable leap in AI-driven vulnerability research, with the ability to uncover thousands of serious flaws across major operating systems and browsers. |
SiliconANGLE - Yutori launches Delegate to turn AI agents into proactive web workers | Delegate is an eager AI agent that the company says never sleeps and is always prepared, at the user’s beck and call, to take whatever the user needs off their plate. Readers are already being inundated with explanations of agentic AI, the next stage after chatbots: systems that sit and wait for commands, while agents can wake up and run on schedule, coordinate, organize and think ahead. |
ZDNet - Google brings Auto Browse and Skills to Chrome Enterprise - and a new 'Gemini Summary' | One of the biggest changes coming to Google's workplace browser, Chrome Enterprise, is a Gemini feature called Auto Browse. I already tested Auto Browse in the consumer version of Chrome, but now it's available to Workspace users in the US. It adds autonomous, agent-like capabilities that can take action on your behalf based on what's open in your browser. |
Computerworld - Claude Mythos signals a new era in AI-driven security, finding 271 flaws in Firefox | The Claude Mythos Preview appears to be living up to the hype, at least from a cybersecurity standpoint. The model, which Anthropic rolled out to a small group of users, including Firefox developer Mozilla, earlier this month, has discovered 271 vulnerabilities in version 148 of the browser. All have been fixed in this week’s release of Firefox 150, Mozilla emphasized. |
The reason? It's what customers around the world, wary of centralized, proprietary, and US-centric tech services and hyperclouds, want. SUSE research revealed at the company's annual trade show in Prague, Czech Republic, that among IT leaders worldwide, 98% prioritize digital sovereignty. Of those, just over half (52%) are already taking steps to achieve freedom from US-based hypercloud providers such as Microsoft and Google. | |
To that end, it has announced a major revamp of Gemini Enterprise, saying it intends to transition AI from an isolated productivity tool into a “secure, collaborative autonomous engine” for business. “Companies are ready to build their agentic task force, but this demands doing so within a secure and governed environment,” the company said. “This includes creating and deploying agents with their own identity, registry, and gateway so they can always be traced, monitored, and managed.” | |
Computerworld - Microsoft trims cloud desktop pricing, even as it boosts AI costs | First, Microsoft has reduced pricing for Windows 365 and AVD in select configurations by 20%. In particular, the company is slashing prices for persistent desktop deployments and lower-tier virtual machines (VMs). These are the instances commonly used by task workers and call centers, not by developers or white-collar office worker bees. |
ZDNet - Scaling agentic AI demands a strong data foundation - 4 steps to take first | According to IDC, by 2026, 40% of all Global 2000 job roles will involve working with AI agents, redefining long-held traditional entry, mid, and senior level positions. But the journey will not be smooth. By 2027, companies that do not prioritize high-quality, AI-ready data will struggle to scale generative AI and agentic solutions, resulting in a 15% loss in productivity. |
Computerworld - Gemini Enterprise update brings AI agents into collaborative workflows | Among these are new ways to interact with Gemini Enterprise, either independently or alongside colleagues. Projects is a new collaborative workspace that enables teams to interact with a shared “expert” chatbot that’s connected to specific data sources from Google Workspace, Microsoft 365 and team chats. |
Development of agentic AI now spans many business domains. According to Anthropic, a provider of large language models (LLMs), AI agents are most commonly deployed in software engineering, accounting for roughly half of use cases, followed by back-office automation, marketing, sales, finance, and data analysis. | |
For enterprise AI application developers who are training their own models, this research provides a proven blueprint for maximizing return on investment. It shows that AI reasoning does not necessarily require spending huge amounts on frontier models. Instead, smaller models can yield stronger performance on complex tasks while keeping per-query inference costs manageable within real-world deployment budgets. | |
This is not to say that cloud computing is inherently unstable or that its advantages—agility, scalability, rapid deployment—are a mirage. Enterprises aren’t abandoning the cloud. Far from it. Adoption continues at pace, even as these high-profile outages occur. The question is not whether the cloud is worth it, but rather, how much unreliability is acceptable for all that innovation and efficiency? | |
Computerworld - World ID expands its ‘proof of human’ vision for the AI era | World ID is taking a unique (and to some, controversial) approach to this challenge by building a ‘digital proof of human’ ecosystem for the internet. Today, at its “Lift Off” event, the Sam Altman co-founded initiative made a series of announcements, which included the launch of version 4.0 of its World ID protocol, a World ID app, World ID for Business, World ID for Agents, a new verification tool called Selfie Check, new monetization programs, and integrations with Zoom and Okta. |
SiliconANGLE - As AI powers Google, what’s next for Google Cloud | This is where we believe Google LLC has an underappreciated advantage. Our research indicates the architectures that will win in the agentic era are the ones that behave like an end-to-end system – where the model, the cognitive engine and the infrastructure are tightly integrated, operate within a single trusted boundary, and enforce consistent security controls without turning economics into a tax at large scale. That is the premise behind our Google thesis. Google has decades of infrastructure and data engineering excellence, and we think it is well-positioned to evolve its cloud from a reactive system of intelligence into an environment that can execute in real time, at scale, with durability. |
Available for download or accessible through APIs, it works by having enterprises define a curated “search space” of approved reports, documents, or application endpoints paired with metadata, and then using vector-based similarity to match a user’s natural language query to the most relevant of pre-approved target, said Tirthankar Lahiri, SVP of mission-critical data and AI engines at Oracle. | |
The reason is simple: the skills that matter most once AI enters real workflows have less to do with interacting with tools and more to do with judgment. The durable capabilities emerging in the AI era include output validation, data literacy, process understanding, and the ability to challenge automated recommendations. Tool-specific skills, by contrast, tend to age quickly as models and interfaces evolve. | |
There are multiple ways to measure the speed of an LLM. The time to first token, or TTFT, is important for real-time, interactive applications where the end user will be daydreaming while waiting for some answer on the screen. Some models start the response faster, but then poke along. Others take longer to begin responding. If you’re going to be using the LLM in the background or as a batch job, this number isn’t as important. | |
In its update today, Anthropic revealed new “organization-wide controls” to help corporate teams deploy its autonomous Claude Cowork service. Meanwhile, OpenAI took a different path, slashing the cost of the “Pro” subscription to access its popular Codex programming tools. | |
ZDNet - I asked 5 data leaders about how they use AI to automate - and end integration nightmares | "We're building AI capability to go out, bring the data back, put it into the database, find out when it's not right, and point out illogical elements, such as where the meter reading is the same as last month, so there must be something wrong," he said. |
To ease the transition, Anthropic offered each subscriber a one-time credit equal to their monthly subscription price, redeemable by April 17 and valid for 90 days across Claude Code, Claude Cowork, chat, or connected third-party tools. The company also introduced pre-purchase extra usage bundles at discounts of up to 30% for subscribers who want to continue running OpenClaw with Claude as the underlying model. | |
Information Week - A practical guide to controlling AI agent costs before they spiral | For AI agents, non-determinism has the effect of making it virtually impossible to anticipate exactly how an agent will fulfill a request -- or even to assume that the way it completed a task historically will continue to be the way it does so in the future. By extension, token costs, infrastructure resource consumption rates and agent maintenance requirements may also vary. |
SiliconANGLE - Identity theft becomes the new perimeter as attackers bypass security defenses | In 2026, attackers are now abusing valid credentials and trusted integrations to move through systems undetected instead of relying on malware or exploiting software vulnerabilities. A key driver of the trend is a significant rise in infostealer malware, including families such as LummaC2. Infostealer tools harvest browser-stored passwords, session cookies and authentication tokens before packaging them into data sets that are sold on underground marketplaces to other threat actors. |
Interviews with analysts, CIOs, and AI platform and governance leaders point to a consistent pattern. The problem is not that AI fails technically. It’s that enterprises are applying legacy budgeting, operating, and accountability models to a technology whose economics behave very differently. As a result, ROI erodes not because AI stops working, but because organizations lose the ability to explain, defend, and prioritize it. | |
Codex is a coding assistant offered as part of ChatGPT that allows developers to interact directly with code repositories by issuing prompts that trigger automated tasks such as code generation, reviews and pull requests. The tasks run inside managed container environments that clone repositories and authenticate using short-lived GitHub OAuth tokens, creating a useful but sensitive execution layer. | |
For cybersecurity leaders, this lesson is particularly relevant. Many organizations are currently evaluating AI as a tool to incrementally improve existing workflows—automating alerts, accelerating triage, or enhancing reporting. While these gains are valuable, they are not transformative. The real risk lies in failing to recognize how AI may fundamentally alter the nature of threats, attack surfaces, and defensive strategies. | |
However, the next generation of digital systems increasingly interacts with regional regulations, real-time decision loops, and the physical world in general. These factors do not tolerate distance well. Smart traffic systems can’t wait for a round-trip to distant cloud regions. Industrial control systems can’t halt operations because a wide-area link is congested. AI-driven video analytics becomes costly and inefficient when every frame must be sent back to a centralized platform for inference. In these environments, it matters where the data is created and processed and where decisions are made. | |
Computerworld - AI regulations are already out of date — IT leaders need to think ahead | Pope said something similar during the AI in the Age of Regulation discussion. No federal regulations will come out of the United States anytime soon, and states are publishing their own rules, she said. California, for instance, is focused on transparency, watermarking, and how AI will affect individuals and groups. |
InfoWorld - Anthropic throttles Claude subscriptions to meet capacity | The rationale here is that by accelerating how quickly users hit their session limits within these windows, Anthropic is effectively redistributing access to prevent system overloads while still preserving overall weekly usage quotas. |
ZDNet - 5 security tactics your business can't get wrong in the age of AI - and why they're critical | Panayi said the multifaceted nature of AI cybersecurity means professionals should expect new roles and responsibilities to emerge, with people sharing knowledge and swapping between teams to create a more powerful approach. |
VentureBeat - Oracle converges the AI data stack to give enterprise agents a single version of truth | Enterprise data teams moving agentic AI into production are hitting a consistent failure point at the data tier. Agents built across a vector store, a relational database, a graph store and a lakehouse require sync pipelines to keep context current. Under production load, that context goes stale. |
SiliconANGLE - RSAC 2026: AI hype meets operating model reality | Organizations are still struggling to consolidate the sprawl of tools in their security stacks and at the same time apply zero-trust principles. To avoid AI becoming yet another layer, organizations must tie AI to clear outcomes and integrate intelligence into operating processes. Enterprise Technology Research survey data captures the challenge. At least 90% of organizations say they’re leveraging AI somewhere in their security stack, but 75% are applying AI to less than 10% of their security portfolio. |
Computerworld - OpenAI’s Foundation play reframes the AI roadmap for IT leaders | It’s an important development that enterprises can learn from, particularly in legacy environments that still hold a ‘don’t share your data’ mindset, Jackson noted. Enterprises are still siloed or protective of their data, and often don’t look at it holistically, even across internal departments. |
Strong winds are buffeting the tightrope and there is legitimate concern that protection of global networks could fall into the digital abyss. At this week’s gathering of cybersecurity professionals for RSAC in San Francisco, a steady drumbeat of keynotes and side sessions offered evidence that threat actors have not only adopted AI, but they are having success in using autonomous technology to fuel identity-based attacks, large-scale denials of service, and poisoning of the software supply chain. | |
Computerworld - Microsoft backtracks on Copilot Chat access in M365 apps | Microsoft is set to remove Copilot Chat access within Microsoft 365 apps such as Word, Excel, and PowerPoint for large M365 commercial customers starting April 15 — a “mystifying backtrack,” according to one technology industry analyst. |
SiliconANGLE - The $1T infrastructure war: How Nvidia is replatforming the agentic era | The Agent Toolkit isn’t about owning the “intelligence” (the AI models); it’s about owning the infrastructure beneath every enterprise agent. Whether you’re running GPT-4, Claude or Llama, Nvidia wants to be the plumbing. Not the brain — the substrate. |
Computerworld - Google targets AI inference bottlenecks with TurboQuant | By compressing these workloads more aggressively without affecting output quality, TurboQuant could allow developers to run more inference jobs on existing hardware and ease some of the cost pressure around deploying large models. |
Computerworld - HP will cram a 20-billion-parameter AI model into new AI PCs | Initial HP IQ on-device AI experiences include Ask IQ, which responds to both text and voice inputs; Analyze, which looks at personal files and generates summaries and actionable insights; Notes and Knowledge, which keeps track of interactions and organizes notes; and Meeting Agent, which records notes or captures ideas during meetings. HP says additional capabilities will roll out later this year. |
Forbes - Why AI Cyberattacks Have Made Your Software Security Strategy Obsolete | AI now enables adversaries to scan millions of lines of code in minutes, identify exploitable patterns, generate novel attack vectors, and adapt in real time when initial attempts fail. Phishing campaigns that once required weeks of social engineering now deploy at scale with personalized precision |
SiliconANGLE - Emma Technologies unifies cloud infrastructure governance for legacy IT environments | The challenge of bringing all of these brownfield environments into a unified management plane is immense, because everything has to be rebuilt from scratch, which requires enormous resources. As a result, many enterprises choose to ignore governance, resulting in fragmented visibility over their sprawling infrastructure estates. |
VentureBeat - Cloudflare’s new Dynamic Workers ditch containers to run AI agent code 100x faster | For enterprise technical decision makers, that is the bigger story. Cloudflare is trying to turn sandboxing itself into a strategic layer in the AI stack. If agents increasingly generate small pieces of code on the fly to retrieve data, transform files, call services or automate workflows, then the economics and safety of the runtime matter almost as much as the capabilities of the model. Cloudflare’s pitch is that containers and microVMs remain useful, but they are too heavy for a future where millions of users may each have one or more agents writing and executing code constantly. |
ZDNet - 5 ways to harden your network against the new speed of AI attacks | Modern enterprise networks are widely distributed and can hand off tasks to partners via software-as-a-service. The bad guys are doing the same thing, Mandiant reports, using a "division of labor" model, in which one group uses low-impact techniques like malicious advertisements or fake browser updates to gain access to a network, then handing off the compromised target to a secondary group for hands-on access. |
MolmoWebMix, the accompanying dataset, includes 30,000 human task trajectories across more than 1,100 websites, 590,000 individual subtask demonstrations and 2.2 million screenshot question-answer pairs — which Ai2 describes as the largest publicly released collection of human web-task execution ever assembled. | |
If you look at Microsoft as a collection of product lines, it is easy to conclude that Windows 11 and Azure occupy different universes. One is a client operating system that has irritated its users, confused administrators, and pushed hardware refresh cycles in ways many customers did not want. The other is a hyperscale cloud platform selling compute, storage, data services, and AI infrastructure to enterprises. On paper, these are different businesses. In practice, they are part of the same trust system. | |
InfoWorld - New ‘StoatWaffle’ malware auto‑executes attacks on developers | According to NTT Security findings, the malware marks an evolution from the long-running campaign’s user-triggered execution to a near-frictionless compromise embedded directly in developer workflows. Attackers are using blockchain-themed project repositories as decoys, embedding a malicious VS Code configuration file that triggers code execution when the folder is opened and trusted by the victim. |
SiliconANGLE - Solink upgrades VerifEye platform to streamline global security operations centers | The platform centralizes data across departments to allow organizations to monitor access control, detect suspicious transaction patterns and identify operational disruptions such as service delays or unusual traffic flows. Operator actions are also recorded in an audit trail to support compliance and insurance requirements.. |
Computerworld - Zoom sees human conversation as its edge in the agentic AI era | As AI agents become better at acting autonomously on behalf of users, human interactions could shift away from applications like Zoom to those agents. In that scenario, collaboration software apps risk becoming the underlying infrastructure rather than the primary interface, a shift that recently prompted concerns about a broader “SaaS-pocalypse” following the launch of AI agent tools such as Anthropic’s Claude Cowork. |
ZDNet - 3 ways Cisco's DefenseClaw aims to make agentic AI safer | DefenseClaw is the "operational layer" for agentic security that has been missing, according to Cisco's head of AI software, DJ Sampath. It is a tool for oversight that will "keep a claw governed," he wrote in a blog post. "That's zero to governed claw in under five minutes." |
If your docs are not controlled in code, how can you automate them? Sure, AI Helps, but AI is no so good at generating Code, or rather, LLMs are so good at generating text that not taking advantage of this paradigm is a mistake. | |
ZDNet - A chief AI officer is no longer enough - why your business needs a 'magician' too | There's a lot of debate about who should be responsible for ensuring the business makes the most out of generative AI. Some experts suggest the CIO should oversee this crucial role, while others believe the responsibility should lie with a chief data officer. |
AI Supremacy Substack - Cursor's Wild Trajectory to being a Vibe Working Leader | Cursor was founded in 2022 by 4 MIT graduates—Michael Truell, Aman Sanger, Sualeh Asif, and Arvid Lunnemark. Now in 2026, Cursor makes about 60% of its revenue from Enterprise customers where large engineering organizations (like those at Nvidia, Uber, and Shopify) transitioned from pilot programs to full-scale deployments. Anysphere is just one year younger than Anthropic and by far the most promising AI coding startup getting into Enterprise autonomous agents. |
TechTalks Substack - AI won't kill SaaS, but major shifts are coming | A closer look at the very companies building these revolutionary models reveals a different reality. The leading artificial intelligence laboratories still rely heavily on established SaaS products to run their daily operations (both CEOs of OpenAI and Anthropic have been on record saying their organization uses Slack). They have access to the most advanced code-generation tools on the planet, yet they continue to pay for off-the-shelf software. They do this because enterprise software involves much more than generating a functional user interface. |
ZDNet - Is your AI agent a security risk? NanoClaw wants to put it in a virtual cage | This will be the first time a claw-based AI agent can be deployed in this manner, and according to the two organizations, it will take only one command to launch. If a user summons NanoClaw, each agent task is isolated in a Docker container running with Docker Sandboxes. |
Emerging from an open beta, the tool utilizes a "dynamic pruning algorithm" to maintain context in large codebases while scaling output to enterprise complexity. Co-founded by Kiran and Mihir Chintawar in 2024, the company aims to bridge the global engineering shortage by positioning Slate as a collaborative tool for the "next 20 million engineers" rather than a replacement for human developers. | |
While Amazon Bedrock helps you build and scale generative AI applications, Amazon Bedrock AgentCore provides an enterprise-grade infrastructure and operations layer for deploying and managing AI agents at scale. AgentCore itself is completely agnostic about models, frameworks, and integrations, although its starter kit CLI only supports the most prominent of these. | |
Computerworld - Data mining? Old servers could become new source of rare earths | For enterprises themselves, he added, “the implications are primarily economic and operational rather than geopolitical. The ability to capture value from retired hardware depends heavily on how organizations manage the end of life phase of their infrastructure lifecycle. Many companies still treat hardware retirement as a simple disposal exercise. Mixed equipment is often shipped to recyclers with little separation between different component types. In those scenarios most of the recoverable value disappears.” |
The new offering is built natively into the NinjaOne platform and brings together artificial intelligence-driven real-time vulnerability assessment, patch confidence scoring and remediation to allow organizations to proactively fix vulnerabilities, minimize mean time to remediate, and reduce time spent vulnerable. | |
Computerworld - Amazon finds out AI programming isn’t all it’s cracked up to be | The root cause was that AI was effectively treated as an extension of a human operator and granted operator‑level permissions. That’s just stupid. You never give someone —or something — system administration privileges unless they absolutely need it and you completely trust them. Neither was true in this case. So, it was that this combination of high privileges and no supervision blew up. |
ZDNet - After using MacBook Neo, it's clear Windows needs to rethink its PC strategy (and fast) | Apple's new MacBook Neo inserts a wedge into the budget laptop market. It's a product category traditionally dominated by Windows PCs, and Microsoft has been quite comfortable in this space for a long time -- its only real competitor being Chromebooks. |
The enterprise launch arrives barely two weeks after Computer debuted for consumers, where it triggered what the company describes as a viral moment: users on social media demonstrated the agent building Bloomberg Terminal-style financial dashboards, replacing six-figure marketing tool stacks in a single weekend, and automating workflows that previously required dedicated teams. Perplexity says more than 100 enterprise customers messaged the company over a single weekend demanding access. | |
Released today, Version 8.7 focuses on reducing the operational burden of large-scale file environments while giving information technology teams greater control over distributed file systems that support global collaboration, the company said. The target market is organizations whose intellectual property resides in large project files, initially architecture, engineering and construction firms. | |
Computerworld - Zoom expands agentic AI platform to automate enterprise workflows | Zoom also introduced new capabilities across its communications and customer experience tools, including Zoom Phone Mobile, SMS support for the Zoom Virtual Agent AI Receptionist, AI Expert Assist 3.0 for its contact center platform, natural-language workflow orchestration for customer interactions, and new meeting security enhancements. |
SiliconANGLE - ORO Labs raises $100M to expand procurement orchestration platform | The platform provides a centralized intake and orchestration layer that connects employees, procurement teams, finance systems and suppliers, allowing organizations to route purchasing requests, approvals and compliance checks through a single workflow framework. ORO’s platform is designed to manage procurement processes across distributed enterprise environments while maintaining policy enforcement and audit tracking. |
Computerworld - Storage vendor offers a real guarantee — but check out those fine-print exceptions | Let’s start with the guarantee, which relates to customers using its Artesca storage line: “A $100,000 financial guarantee to customers if an external cyberattack destroys or encrypts data stored immutably on Artesca. The program applies to every Artesca customer without requiring the purchase of additional services. As long as organizations keep Artesca up to date and protect data using Object Lock in compliance mode, they qualify for the guarantee.” |
ZDNet - 5 security tactics your business can't get wrong in the age of AI - and why they're critical | Lovelock told ZDNET that one key issue is that organizations can't yet benefit from access to measurable, definable, and certifiable AI safety, meaning end-user security requirements are unlikely to be met by many of their providers. |
Computerworld - It looks like Macs are becoming the value option | As a result, the number of people Apple can offer a Mac to is growing as rapidly as the product matrix. Future Ultra Macs will take that reach all the way up to the very, very top tiers currently served by furiously expensive PC workstations, while the Neo range (which I’m willing to bet gets a backlit keyboard and more memory next year) extends its hand all the way to students and general purpose computer users. |
VentureBeat - Anthropic and OpenAI just exposed SAST's structural blind spot with free tools | OpenAI launched Codex Security on March 6, entering the application security market that Anthropic had disrupted 14 days earlier with Claude Code Security. Both scanners use LLM reasoning instead of pattern matching. Both proved that traditional static application security testing (SAST) tools are structurally blind to entire vulnerability classes. The enterprise security stack is caught in the middle. |
The search giant and, increasingly, AI leader today announced a sweeping series of updates to its Gemini AI models embedded into Google Workspace — the productivity suite of cloud-based apps including Drive, Docs, Sheets, Slides, and more. They're being made available both to individual consumers and enterprises, though you'll need an AI Pro ($20 per month) or higher subscription plan for the former, and your enterprise will need to be enrolled in the "Gemini Alpha" program and have the features switched on by an administrator. | |
InfoWorld - Anthropic debuts Claude Marketplace to target AI procurement bottlenecks | Called Claude Marketplace, the platform currently has a limited set of partners, including Replit, Lovable Labs, GitLab, Snowflake, Harvey AI, and Rogo, offering tools across software development, legal workflows, financial analysis, and enterprise data operations, respectively. |
SiliconANGLE - Mend.io launches AI system prompt hardening solution to secure LLM instructions | Mend.io said its new system prompt hardening capability helps move security teams beyond ad hoc testing and manual red teaming to test LLM responses to attacks in a standardized framework for managing security. |
Computerworld - M365 Copilot gets its own version of Claude Cowork | The Microsoft 365 Copilot Wave 3 brings new agentic AI tools to create and edit documents, alongside the launch of an E7 price tier that bundles AI tools with M365 apps for $99 per user each month. Businesses should be wary of the limitations and risks related tousing Copilot Cowork, say analysts. |
SiliconANGLE - Exclusive: Virtana customizes its observability platform for AI workloads | The platform combines application telemetry with infrastructure-level data to automatically correlate performance issues across hybrid environments. Virtana said its approach identifies root causes more quickly and supports what it calls “system-level observability” rather than the code-centric monitoring used by many legacy APM platforms. |
ZDNet - AI is supercharging cloud cyberattacks - and third-party software is the most vulnerable | The report concludes that the best way to fight AI-powered attacks is with AI-augmented defenses: "This activity, along with AI-assisted attempts to probe targets for information and continued threat actor emphasis on data-focused theft, indicates that organizations should be turning to more automatic defenses." |
InfoWorld - How generative UI cut our development time from months to weeks | I lead and implemented an approach that exists somewhere in between. We specify a library of components and allowable layout patterns that define the constraints of our design system. The AI then chooses components from this library, customizes them based on context and lays them out appropriately for each unique user interaction. |
SiliconANGLE - Google enhances Docs, Sheets, Slides and Drive with deeper Gemini integration | With this recent update, Google introduced a new “Help me create” experience in Docs. Users can describe what they want to create and it will follow instructions and synthesize information by looking over Drive, Gmail, Chat and web sources to generate a fully formed draft. |
Computerworld - Apple’s new $599 MacBook Neo is a nightmare for Windows OEMs | “A watershed event,” said Asymco analyst Horace Dediu. “First Mac with a mobile processor and the end of the disruptive arc of mobile computing. From Motorola to Intel to Apple silicon M, now personal computing is an accessory to mobile computing. A sharp punctuation point.” |
SiliconANGLE - With $200M in funding, Eridu wants to break through the network wall holding back AI | Perkins said that the bandwidth, latency, power consumption and radix (the number of input/output ports) of existing network switches are tightly coupled to the silicon architecture they’re based on, which was designed for cloud data centers that are much smaller than today’s emerging AI factories. “This silicon architecture has fundamentally been the same for the last two decades and is only incrementally improved with a doubling of capacity every two to 2.5 years,” he said. “We believe that these incremental improvements leave a lot of performance on the table.” |
ZDNet - Why AI is both a curse and a blessing to open-source software - according to developers | At FOSDEM 2026 in Brussels, Belgium, Stenberg said that, until early 2025, roughly one in six security reports to cURL were valid. That's because, "in the old days, you know, someone actually invested a lot of time [in] the security report. There was a built-in friction here, but now there's no effort at all in doing this. The floodgates are open. Send it over." |
So, why the total absence of trust? Here’s the bad news. On the back of AI, cybercrime has become a global superpower, with an estimated $10.5 trillion coming from extortion, phishing, hacks, and ransomware – by my calculations, that is fifteen times the value of the global AI market. | |
Forbes - AI Agents Now Buy From Other AI Agents — What Leaders Must Know | This shift is already reshaping enterprise procurement, logistics and consumer planning at scale. Agentic systems operate in layered pipelines where one model's output becomes another model's input. When your planning agent selects a florist, it may already be transacting through a vendor agent that has pre-negotiated pricing with a supplier agent upstream. The speed and autonomy are extraordinary. The accountability gap, however, is just as significant. |
VentureBeat - Pentagon vendor cutoff exposes the AI dependency map most enterprises never built | A January 2026 Panorays survey of 200 U.S. CISOs put a number on the problem: Only 15% said they have full visibility into their software supply chains, up from just 3% a year ago. And 49% had adopted AI tools without employer approval, according to a BlackFog survey of 2,000 workers at companies with more than 500 employees; 69% of C-suite members said they were fine with it. |
Confluent's latest Confluent Intelligence features include support for both Anthropic's Model Context Protocol (MCP) and the Agent2Agent (A2A) protocol within Streaming Agents, plus a new multivariate anomaly detection capability. All technically credible additions. But the more difficult challenge is about whether enterprises have the data infrastructure, governance maturity, and organizational readiness to make agent coordination actually work. | |
The Juno platform was originally developed as a threat hunting platform capable of analyzing activity across both cloud-native and on-premises environments but is now being positioned as a broader strategic cybersecurity assistant. The platform analyzes telemetry from cloud infrastructure, containers and endpoints to help security teams identify threats, investigate incidents and understand attack paths across complex enterprise environments. | |
VentureBeat - Databricks built a RAG agent it says can handle every kind of enterprise search | Databricks set out to fix that with KARL, short for Knowledge Agents via Reinforcement Learning. The company trained an agent across six distinct enterprise search behaviors simultaneously using a new reinforcement learning algorithm. The result, the company claims, is a model that matches Claude Opus 4.6 on a purpose-built benchmark at 33% lower cost per query and 47% lower latency, trained entirely on synthetic data the agent generated itself with no human labeling required. That comparison is based on KARLBench, which Databricks built to evaluate enterprise search behaviors. |
Computerworld - Apple’s new $599 MacBook Neo is a nightmare for Windows OEMs | The company is openly targeting customers who want to shift from Windows to a better operating system with the hardware to match. A visit to the product pages on the Apple website offers a “Switch from PC to Mac” section where you’ll find help and answers to decide if the time is right to upgrade to Mac. |
ZDNet - The biggest AI threats come from within - 12 ways to defend your organization | Like Thor and Loki, or Batman and the Joker, the two foes constantly have to outpace and outmaneuver one another in what's shaping up to be a long, possibly never-ending arms race. (On a related note, AI developers like OpenAI have their own security arms race to contend with: the better that their models can protect against prompt injection attacks, the more cunning those attacks become.) |
SiliconANGLE - Agentic business intelligence startup WisdomAI shifts from insights to action | The company says it’s using artificial intelligence agents to tackle the “last mile” problem inherent in almost every modern data stack: translating that data into decisions. Until now, the decision-making has always been done by humans, who still have to switch between Excel spreadsheets and BI dashboards to gather all of the information they need to know what to do |
Computerworld - How vibe coding is reshaping software development, and what it breaks along the way | AI-powered “vibe coding” is moving from experimentation to real production software. But as developers and AI agents begin building side by side, enterprises face new questions around quality control, tech debt, team structure, and the future of junior engineers |
ZDNet - Will AI make cybersecurity obsolete or is Silicon Valley confabulating again? | To the rescue come the major creators of AI models, OpenAI, Anthropic, and Google. All three offer tools that could mitigate failures and security breaches in LLMs and the agentic programs built on top of them. |
InfoWorld - The right way to architect modern web applications | We saw it in the early 2000s, when server-rendered, monolithic applications were the default. We saw it again in the late 2000s and early 2010s, when the industry pushed aggressively toward rich client-side applications. And we saw it most clearly during the rise of single-page applications, which promised desktop-like interactivity in the browser but often delivered something else entirely: multi-megabyte JavaScript bundles, blank loading screens, and years of SEO workarounds just to make pages discoverable. |
Computerworld - What is digital employee experience — and why is it more important than ever? | Digital employee experience is a measure of how workers perceive and interact with the many digital tools and services they use in the workplace. It examines how employees feel about these technologies, including systems, software, and devices. |
Compensation follows leverage. In major markets, it’s common to see total annual compensation for experienced cloud architects exceed $200,000, particularly when the role includes broad platform scope, security accountability, and cross-domain influence. One good architect can keep a large organization out of trouble in ways that save far more than the cost of the role. | |
Computerworld - 3 Android theft protection additions you should absolutely activate | More than anything, though, no naughty Android app can just magically plop itself onto your phone and then access private info. Apps only appear if you explicitly install ’em — and even then, they’re only able to access sensitive data and areas of your device if you approve the permissions to permit that. |
ZDNet - Why enterprise AI agents could become the ultimate insider threat | Generative AI is moving from chatbot to autonomous actor. When agents can launch other agents, spend money, and modify systems, the line between productivity tool and insider threat disappears. |
SiliconANGLE - DeepKeep launches AI agent attack surface scanner to map enterprise risk | The release today includes AI Agent Scanner, which provides immediate visibility into what AI agents can access, which tools and data they interact with and where potential vulnerabilities exist to meet a pressing enterprise need as the AI agent attack surface grows. The solution performs robust attack surface scanning to map an agent’s entire threat landscape, identifying connected tools and their intents, data sources and potential vulnerabilities. |
VentureBeat - When AI lies: The rise of alignment faking in autonomous systems | Alignment faking usually happens when earlier training conflicts with new training adjustments. AI is typically “rewarded” when it performs tasks accurately. If the training changes, it may believe it will be “punished” if it does not comply with the original training. Therefore, it tricks developers into thinking it is performing the task in the required new way, but it will not actually do so during deployment. Any large language model (LLM) is capable of alignment faking. |
Computerworld - OAuth phishers make ‘check where the link points’ advice ineffective | Microsoft has warned that phishers are exploiting a built-in behavior of the OAuth authentication protocol to redirect victims to malware, using links that point to legitimate identity provider domains such as Microsoft Entra ID and Google Workspace. The links look safe but ultimately lead somewhere that isn’t. |
ZDNet - Why encrypted backups may fail in an AI-driven ransomware era | Think your encrypted backups are safe? AI-driven ransomware now infiltrates networks, corrupts recovery points, and silently targets backup systems before you ever realize your data protection strategy has failed. |
SiliconANGLE - Cloudflare warns AI and SaaS integrations are fueling industrial-scale cybercrime | The report provides various examples to back up its claims. In a campaign tracked as GRUB1, attackers compromised a trusted SaaS-to-SaaS connection and then used generative artificial intelligence to navigate complex enterprise platforms in real time. The actor turned a single integration into a multitenant breach with supply chain implications by identifying high-value database tables moments before accessing production environments. |
As I’ve argued, the real enterprise AI challenge is no longer training. It’s inference: applying models continuously to governed enterprise data, under real-world latency, security, and cost constraints. That shift matters because once inference becomes the steady-state workload of the enterprise, infrastructure that once seemed necessary but dull suddenly becomes strategic. | |
ZDNet - Rolling out AI? 5 security tactics your business can't get wrong - and why | The same capabilities that make AI useful also make it exploitable. In fact, the rate at which emerging technologies are advancing intensifies that uncomfortable reality by the minute. |
Information Week - Who really sets AI guardrails? How CIOs can shape AI governance policy | Somewhere between the requirements of government policy, the terms set by the vendor, the pressure of the customer and the guidance of the board, CIOs must chart a path that maximizes AI utility while protecting the business. While they cannot dictate the environment, they can make critical choices within it. |
InfoWorld - OpenAI launches stateful AI on AWS, signaling a control plane power shift | The company has announced that it will soon offer a stateful runtime environment in partnership with Amazon, built to simplify the process of getting AI agents into production. It will run natively on Amazon Bedrock, be tailored for agentic workflows, and optimized for AWS infrastructure. |
Out of the initial chaos came a clear lesson about the role of an AI coder. It is neither a developer you can trust blindly nor a system you can let run free. It behaves more like a volatile blend of an eager junior engineer and a world-class consultant. Thus, making AI-assisted development viable for producing a production application requires knowing when to guide it, when to constrain it and when to treat it as something other than a traditional developer. | |
InfoWorld - The browser is your database: Local-first comes of age | But an alternative is emerging. The idea is to embed a relational database directly in the browser, with a slice of the data, and let a synchronization (sync) engine keep everything consistent. The browser interacts with a local datastore that is synced to the server in the background. This means instant interactivity on the front end while maintaining symmetry with the back end. This next-generation browser has a more resilient state-of-record, not just a temporary cache. |
SiliconANGLE - Figma’s orchestration bet: Why MCP network effects redefine software defensibility | Figma isn’t just a design tool anymore. It’s a shared design system across engineering, product, marketing and increasingly nondesigners. Nearly 60% of Figma Make files are now created by nondesigners. More than 75% of customers use multiple Figma products. |
The Coinerella approach is to deliberately refuse to let the platform drift toward AWS and US-based hyperscalers, driven by practical considerations such as data residency, General Data Protection Regulation (GDPR) compliance, reducing concentration risk, and demonstrating the operational viability of European infrastructure. Leaders often talk about sovereignty until the first production incident, the first compliance review, or the first integration gap. Coinerella remains committed and is addressing the consequences. | |
ZDNet - Is Microsoft really spying on you with Windows telemetry? | But you know what? More than a decade later, people are still spreading those conspiracy theories. Microsoft is spying on you! Redmond is collecting mountains of personal data and using it for ... advertising, I guess? And the rise of AI means that there are even bigger rabbit holes to go down. |
And yes, I know, nobody is doing 1,800 meaningful commits. But that’s the point. The metric is already being gamed, and agents make gaming effortless. If your organization starts celebrating “commit velocity” in the agent era, you are not measuring productivity. You are measuring how quickly your team can manufacture liability. | |
Computerworld - US orders diplomats to push back on data sovereignty | At the same time, support for data sovereignty is growing, especially in Europe, where there are concerns about privacy, surveillance, and US dominance in AI and tech. The EU’s GDPR is mentioned in the document as an example of rules that the US considers unnecessarily restrictive. |
InfoWorld - 7 ways to tame multicloud chaos with generative AI | Standardizing on a single cloud infrastructure is much easier than pursuing a multicloud strategy. In a single-cloud environment, IT leaders can optimize skill sets, centralize data more easily, secure infrastructure with fewer tools, and gain many other operational benefits. Yet 89% of enterprises report they are pivoting to multicloud adoption. Reasons for choosing to operate across multiple clouds include mitigating risk, reducing service interruptions, and avoiding vendor lock-in. |
Which is why — for many of these organizations — the default lens for agents is frequently automation, not agency. Of eliminating people to reduce cost. Of deterministic workflows. Of eliminating rather than enabling judgement. Of detailed operations rather than delegated outcomes. | |
Computerworld - Anthropic targets core business systems with new Claude plug-ins | In a blog post, the company said new connectors are available for widely used enterprise platforms, including Google Workspace tools such as Calendar, Drive, and Gmail, as well as DocuSign, FactSet, MSCI, and LegalZoom, while partners such as Slack, LSEG, and S&P Global have built plug-ins for joint customers. |
Anthropic opened its virtual "Briefing: Enterprise Agents" event on Tuesday with a provocation. Kate Jensen, the company's head of Americas, told viewers that the hype around enterprise AI agents in 2025 "turned out to be mostly premature," with many pilots failing to reach production. "It wasn't a failure of effort, it was a failure of approach, and it's something we heard directly from our customers," Jensen said. | |
Computerworld - After OpenClaw backlash, Quill bets on security-by-design agentic AI | Naturally, though, enterprises and users may be concerned about how they can remain in control of their data. Addressing this, Quill is “local-first with options,” meaning transcription and speaker recognition run on-device and audio never leaves that environment. The agent never stores data, and enterprises have access to configurable endpoints to ensure zero exposure. |
Forbes - AI Rattles Cybersecurity Markets: What Anthropic’s Code Security Actually Does | Anthropic introduced Claude Code Security, an AI driven capability embedded into its Claude Code platform. Within hours, a broad set of cybersecurity equities declined sharply. The prevailing narrative formed quickly: AI is now replacing cybersecurity tools. |
ZDNet - Copilot quietly grabs your data from other Microsoft products now - here's how to opt out | Known as "Microsoft usage data," the setting lets Copilot refer to your data from Bing, MSN, Edge, and other Microsoft products that you've used, as spotted by Windows Latest. Accessible at the Copilot website and through the mobile app, the setting appears to be relatively new and is part of the Memory option in Copilot. This option allows the AI to recall your conversation history, any facts and instructions you share, and certain data from Microsoft products, all in an effort to personalize Copilot. |
Forbes - Anthropic Leans Into Enterprise With Managed Claude Cowork Plugins | Companies can build AI agents that adapt to their unique workflows, using the same tools they already trust, but are private and protected in the enterprise, with admin control. The goal for Anthropic is to move AI from being a peripheral tool to an integrated layer of business operations, where every department gets its own specialized assistant. |
Computerworld - What really caused that AWS outage in December? | The back-story was broken by the Financial Times, which reported the 13-hour outage was caused by a Kiro agentic coding system that decided to improve operations by deleting and then recreating a key environment. |
InfoWorld - Compromised npm package silently installs OpenClaw on developer machines | “I mean, they effectively turned OpenClaw into malware that EDR [endpoint detection and response ] isn’t going to stop,” said David Shipley of Beauceron Security. It is “deviously, terrifyingly brilliant.” |
SiliconANGLE - Veeam launches Agent Commander to tackle AI risk and reverse agent mistakes | Agent Commander addresses what Veeam calls the most critical gap in AI infrastructure today: trust. Veeam argues that as AI agents scale, data risk and AI risk have become the same problem and that an agent is only as trustworthy as the data it can see, access and act on. Added to the mix is that sensitive data is being fed into models and acted upon in ways no one approved nor is tracking. |
Our team wanted to get at the root of what IT leaders are thinking, so we surveyed CIOs and CTOs from around the globe. What we found was fascinating: 84% of IT leaders believe automation (the structured, governed execution of repetitive business processes that can be optimized to run without human intervention) must come first if AI is to succeed. Those with mature automation programs were more than twice as likely to describe their AI initiatives as transformational compared to peers still in the early stages. | |
The great promise of generative artificial intelligence was that it would finally clear our backlogs. Coding agents would churn out boilerplate at superhuman speeds, and teams would finally ship exactly what the business wants. The reality, as we settle into 2026, is far more uncomfortable. Artificial intelligence is not going to save developer productivity because writing code was never the bottleneck in software engineering. The true bottleneck is validation. Integration. Deep system understanding. Generating code without a rigorous validation framework is not engineering. It is simply mass-producing technical debt. | |
The Hacker News - Identity Prioritization isn't a Backlog Problem - It's a Risk Math Problem | In modern enterprises, identity risk is created by a compound of factors: control posture, hygiene, business context, and intent. Any one of these can perhaps be manageable on its own. The real danger is the toxic combination, when multiple weaknesses align and attackers get a clean chain from entry to impact. |
The guardrails Treasure Data built live upstream of the code itself. When any user connects to the CDP through Treasure Code, access control and permission management are inherited directly from the platform. Users can only reach resources they already have permission for. PII cannot be exposed. API keys cannot be surfaced. The system cannot speak disparagingly about a brand or competitor. | |
Computerworld - With ‘Frontier,’ OpenAI hopes to own the enterprise agent stack | Frontier is an “end-to-end” platform designed to help “enterprises build, deploy and manage AI agents,” according to OpenAI. It connects agents to core business systems — such as CRM, ERP, and data warehouses — and centralizes how these agents are configured, monitored and governed. |
Everpure is betting that the primary blocker to enterprise AI value is not model quality, compute access, or organizational readiness - it's data infrastructure. Specifically, it’s data fragmentation, poor governance, lack of provenance, and the inability to make data available at machine speed without violating access controls or sovereignty requirements. | |
VentureBeat - How attackers hit 700 organizations through CX platforms your SOC already approved | CX platforms process billions of unstructured interactions a year: Survey forms, review sites, social feeds, call center transcripts, all flowing into AI engines that trigger automated workflows touching payroll, CRM, and payment systems. No tool in a security operation center leader’s stack inspects what a CX platform’s AI engine is ingesting, and attackers figured this out. They poison the data feeding it, and the AI does the damage for them |
InfoWorld - T mistakes that escalate into serious cyber-risk | "A lot of companies go sideways because they don't appreciate the level of risk that lies with IT," Lyborg said. "If they don't do the basics of identity management, access and audit reviews, they can't see or react when something [is off]." |
VentureBeat - Shadow mode, drift alerts and audit logs: Inside the modern audit loop | Traditional software governance often uses static compliance checklists, quarterly audits and after-the-fact reviews. But this method can't keep up with AI systems that change in real time. A machine learning (ML) model might retrain or drift between quarterly operational syncs. This means that, by the time an issue is discovered, hundreds of bad decisions could already have been made. This can be almost impossible to untangle. |
What does this have to do with the current state of AI adoption? The basic message is that if you have an organization that is essentially greenfield, such as a start-up, they can tap into all the AI promises made for the tech. But the overwhelming majority of enterprises are brownfield and trying to modernize their systems with AI and keep the lights on is a major challenge. |