AI agents hold the promise of automatically moving data between systems and triggering decisions, but in some cases, they can act without a clear record of what, when, and why they undertook their tasks.

That has the potential to create a governance problem, for which IT leaders are ultimately responsible. If an organisation can’t trace an agent’s actions and don’t have proper control over its authority, leaders can’t prove that a system is operating safely or even lawfully to regulators.

That’s an issue set to become more important from August this year, as enforcement of the EU AI Act kicks in. According to the text of the Act, there will be substantial penalties for failures of governance relating to AI, especially when used in high-risk areas such as when personally-identifiable information is processed, or financial operations take place.

The post Agentic AI’s governance challenges under the EU AI Act in 2026 appeared first on AI News.

Anthropic’s most capable AI model has already found thousands of AI cybersecurity vulnerabilities across every major operating system and web browser. The company’s response was not to release it, but to quietly hand it to the organisations responsible for keeping the internet running.

That model is Claude Mythos Preview, and the initiative is called Project Glasswing.

The launch partners include Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, Nvidia, and Palo Alto Networks. 

Beyond that core group, Anthropic has extended access to over 40 additional organisations that build or maintain critical software infrastructure. Anthropic is committing up to US$100 million in usage credits for Mythos Preview across the effort, along with US$4 million in direct donations to open-source security organisations. 

A model that outgrew its own benchmarks

Mythos Preview was not specifically trained for cybersecurity work. Anthropic said the capabilities “emerged as a downstream consequence of general improvements in code, reasoning, and autonomy”, and that the same improvements making the model better at patching vulnerabilities also make it better at exploiting them. 

That last part matters. Mythos Preview has improved to the extent that it mostly saturates existing security benchmarks, forcing Anthropic to shift its focus to novel real-world tasks–specifically, zero-day vulnerabilities. These flaws were previously unknown to the software’s developers. 

Among the findings: a 27-year-old bug in OpenBSD, an operating system known for its strong security posture. In another case, the model fully autonomously identified and exploited a 17-year-old remote code execution vulnerability in FreeBSD–CVE-2026-4747–that allows an unauthenticated user anywhere on the internet to obtain complete control of a server running NFS. No human was involved in the discovery or exploitation after the initial prompt to find the bug. 

Nicholas Carlini from Anthropic’s research team described the model’s ability to chain together vulnerabilities: “This model can create exploits out of three, four, or sometimes five vulnerabilities that in sequence give you some kind of very sophisticated end outcome. I’ve found more bugs in the last couple of weeks than I found in the rest of my life combined.” 

Why is it not being released?

“We do not plan to make Claude Mythos Preview generally available due to its cybersecurity capabilities,” Newton Cheng, Frontier Red Team Cyber Lead at Anthropic, said. “Given the rate of AI progress, it will not be long before such capabilities proliferate, potentially beyond actors who are committed to deploying them safely. The fallout–for economies, public safety, and national security–could be severe.” 

This is not hypothetical. Anthropic had previously disclosed what it described as the first documented case of a cyberattack largely executed by AI–a Chinese state-sponsored group that used AI agents to autonomously infiltrate roughly 30 global targets, with AI handling the majority of tactical operations independently. 

The company has also privately briefed senior US government officials on Mythos Preview’s full capabilities. The intelligence community is now actively weighing how the model could reshape both offensive and defensive hacking operations. 

The open-source problem

One dimension of Project Glasswing that goes beyond the headline coalition: open-source software. Jim Zemlin, CEO of the Linux Foundation, put it plainly: “In the past, security expertise has been a luxury reserved for organisations with large security teams. Open-source maintainers, whose software underpins much of the world’s critical infrastructure, have historically been left to figure out security on their own.”

Anthropic has donated US$2.5 million to Alpha-Omega and OpenSSF through the Linux Foundation, and US$1.5 million to the Apache Software Foundation–giving maintainers of critical open-source codebases access to AI cybersecurity vulnerability scanning at a scale that was previously out of reach.

What comes next

Anthropic says its eventual goal is to deploy Mythos-class models at scale, but only when new safeguards are in place. The company plans to launch new safeguards with an upcoming Claude Opus model first, allowing it to refine them with a model that does not pose the same level of risk as Mythos Preview. 

The competitive picture is already shifting around it. When OpenAI released GPT-5.3-Codex in February, the company called it the first model it had classified as high-capability for cybersecurity tasks under its Preparedness Framework. Anthropic’s move with Glasswing signals that the frontier labs see controlled deployment–not open release–as the emerging standard for models at this capability level.

Whether that standard holds as these capabilities spread further is, at this point, an open question that no single initiative can answer.

See Also: Anthropic’s refusal to arm AI is exactly why the UK wants it

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the Cyber Security & Cloud Expo. Click here for more information.

AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.

The post Anthropic keeps new AI model private after it finds thousands of external vulnerabilities appeared first on AI News.

A survey carried out by OutSystems, The State of AI Development 2026 [email wall], argues that AI has moved into early production phase for many enterprises, primarily inside the IT function.

The survey was based on the responses of 1,879 IT leaders, and warns that adoption of AI is in danger of running ahead of governance and integration. The shortfall is a gap between what IT leaders want agents to do and what their organisations can safely control. The report’s authors urge companies to address the controls or guardrails on AI systems, and also stress the importance of integrating new, AI technology into an organisation’s existing platforms.

OutSystems says 97% of its respondents are exploring some form of agentic strategy, with 49% of them describing their current abilities as “advanced” or “expert.” Nearly half of those surveyed say that over half of agentic AI projects have moved from pilot into production, with Indian companies most successful in implementing the technology: 50% of Indian companies say their AI projects are 51% to 75% successful.

Companies are considering where agents should be deployed first, and under what controls, but although “cost reduction or efficiency gains” is the most cited expectation for AI’s effects, only 22% found their deployments most effective in that regard. Instead, the most effective area gains in a business stemmed from equipping software developers with AI tools described as “generative AI-assisted.”

The report’s geography and sector data show that transitions to AI agentic workflows are unevenly distributed. India stands out as the market with the highest share of users considering themselves “expert”, while many organisations in Australia, Brazil, Germany, the Netherlands, the UK, and the US still identify as intermediate stage users. France and Germany are the most dubious of AI adoption, with Germany recording the highest share of leaders not using agentic AI in any form.

The post AI’s software development success and central management needs appeared first on AI News.

A new open-source toolkit from Microsoft focuses on runtime security to force strict governance onto enterprise AI agents. The release tackles a growing anxiety: autonomous language models are now executing code and hitting corporate networks way faster than traditional policy controls can keep up.

AI integration used to mean conversational interfaces and advisory copilots. Those systems had read-only access to specific datasets, keeping humans strictly in the execution loop. Organisations are currently deploying agentic frameworks that take independent action, wiring these models directly into internal application programming interfaces, cloud storage repositories, and continuous integration pipelines.

When an autonomous agent can read an email, decide to write a script, and push that script to a server, stricter governance is vital. Static code analysis and pre-deployment vulnerability scanning just can’t handle the non-deterministic nature of large language models. One prompt injection attack (or even a basic hallucination) could send an agent to overwrite a database or pull out customer records.

Microsoft’s new toolkit looks at runtime security instead, providing a way to monitor, evaluate, and block actions at the moment the model tries to execute them. It beats relying on prior training or static parameter checks.

Intercepting the tool-calling layer in real time

Looking at the mechanics of agentic tool calling shows how this works. When an enterprise AI agent has to step outside its core neural network to do something like query an inventory system, it generates a command to hit an external tool.

Microsoft’s framework drops a policy enforcement engine right between the language model and the broader corporate network. Every time the agent tries to trigger an outside function, the toolkit grabs the request and checks the intended action against a central set of governance rules. If the action breaks policy (e.g. an agent authorised only to read inventory data tries to fire off a purchase order) the toolkit blocks the API call and logs the event so a human can review it.

Security teams get a verifiable, auditable trail of every single autonomous decision. Developers also win here; they can build complex multi-agent systems without having to hardcode security protocols into every individual model prompt. Security policies get decoupled from the core application logic entirely and are managed at the infrastructure level.

Most legacy systems were never built to talk to non-deterministic software. An old mainframe database or a customised enterprise resource planning suite doesn’t have native defenses against a machine learning model shooting over malformed requests. Microsoft’s toolkit steps in as a protective translation layer. Even if an underlying language model gets compromised by external inputs; the system’s perimeter holds.

Security leaders might wonder why Microsoft decided to release this runtime toolkit under an open-source license. It comes down to how modern software supply chains actually work.

Developers are currently rushing to build autonomous workflows using a massive mix of open-source libraries, frameworks, and third-party models. If Microsoft locked this runtime security feature to its proprietary platforms, development teams would probably just bypass it for faster, unvetted workarounds to hit their deadlines.

Pushing the toolkit out openly means security and governance controls can fit into any technology stack. It doesn’t matter if an organisation runs local open-weight models, leans on competitors like Anthropic, or deploys hybrid architectures.

Setting up an open standard for AI agent security also lets the wider cybersecurity community chip in. Security vendors can stack commercial dashboards and incident response integrations on top of this open foundation, which speeds up the maturity of the whole ecosystem. For businesses, they avoid vendor lock-in but still get a universally scrutinised security baseline.

The next phase of enterprise AI governance

Enterprise governance doesn’t just stop at security; it hits financial and operational oversight too. Autonomous agents run in a continuous loop of reasoning and execution, burning API tokens at every step. Startups and enterprises are already seeing token costs explode when they deploy agentic systems.

Without runtime governance, an agent tasked with looking up a market trend might decide to hit an expensive proprietary database thousands of times before it finishes. Left alone, a badly configured agent caught in a recursive loop can rack up massive cloud computing bills in a few hours.

The runtime toolkit gives teams a way to slap hard limits on token consumption and API call frequency. By setting boundaries on exactly how many actions an agent can take within a specific timeframe, forecasting computing costs gets much easier. It also stops runaway processes from eating up system resources.

A runtime governance layer hands over the quantitative metrics and control mechanisms needed to meet compliance mandates. The days of just trusting model providers to filter out bad outputs are ending. System safety now falls on the infrastructure that actually executes the models’ decisions

Getting a mature governance program off the ground is going to demand tight collaboration between development operations, legal, and security teams. Language models are only scaling up in capability, and the organisations putting strict runtime controls in place today are the only ones who will be equipped to handle the autonomous workflows of tomorrow.

See also: As AI agents take on more tasks, governance becomes a priority

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the Cyber Security & Cloud Expo. Click here for more information.

AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.

The post Microsoft open-source toolkit secures AI agents at runtime appeared first on AI News.

AI systems are starting to move beyond simple responses. In many organisations, AI agents are now being tested to plan tasks, make decisions, and carry out actions with limited human input. It is no longer just about whether a model gives the right answer. It is about what happens when that model is allowed to act.

Autonomous systems need clear boundaries. They need rules that define what they can access, what they are allowed to do, and how their actions are tracked. Without those controls, even well-trained systems can create problems that are hard to detect or reverse.

One company working on this problem is Deloitte. The firm has been developing governance frameworks and advisory approaches to help organisations manage AI systems.

From tools to AI agents

Most AI systems in use today still depend on human prompts. They generate text, analyse data, or make predictions, but a person usually decides what happens next. Agentic AI changes that pattern. These systems can break down a goal into steps, choose actions, and interact with other systems to complete tasks.

That added independence brings new challenges. When a system acts on its own, it may take paths that were not fully expected or use data in ways that were not intended.

Deloitte’s work focuses on helping organisations prepare for these risks. Rather than treating AI as a standalone tool, the firm looks at how it fits into business processes, including how decisions are made and how data flows through systems.

Building governance into the lifecycle

Governance should not be added after deployment. It needs to be built into the full lifecycle of an AI system.

This starts at the design stage. Organisations need to define what a system is allowed to do and where its limits are. This may include setting rules around data use and outlining how the system should respond in uncertain situations.

The next stage is deployment. At this point, governance focuses on access and control, including who can use the system and what it can connect to. Once the system is live, monitoring becomes the main concern. Autonomous systems can change over time as they interact with new data. Without regular checks, they may drift away from their original purpose.

The role of transparency and accountability

As AI systems take on more responsibility, it becomes more difficult to trace how decisions are made. This creates a demand for stronger transparency. Deloitte’s work highlights the importance of keeping track of how systems operate. This includes logging actions and documenting decisions. These records help organisations in determining what happened if something goes wrong. If an autonomous system takes an action, there needs to be clarity about who is responsible.

Research from Deloitte shows that adoption of AI agents is moving faster than the controls needed to manage them. Around 23% of companies already use them, and that figure is expected to reach 74% within two years. Only 21% report having strong safeguards in place to oversee how they behave.

Real-time oversight for AI agents

Once an autonomous system is active, the focus shifts to how it behaves in real-world conditions. Static rules are not always enough, and systems need to be observed as they operate.

Deloitte’s approach includes real-time monitoring, allowing organisations to track what an AI system is doing as it performs tasks. If the system behaves in an unexpected way, teams can step in quickly. This may involve pausing certain actions or adjusting permissions. Real-time oversight also helps with compliance. In regulated industries, companies need to show that systems follow rules and standards.

In practice, these controls are starting to appear in operational settings. Deloitte describes scenarios where AI systems monitor equipment performance across sites. Sensor data can signal early signs of failure, which can trigger maintenance workflows and update internal systems. Governance frameworks define what actions the system can take, when human approval is required, and how decisions are recorded. The process runs across multiple systems, but from a user’s point of view, it appears as a single action.

Governance is part of discussions at AI & Big Data Expo North America 2026, taking place on May 18–19 in Santa Clara, California. Deloitte is listed as a Diamond Sponsor for the event, placing it among the firms contributing to conversations around how autonomous systems are deployed and controlled in practice.

The challenge is not just building smarter systems, but ensuring they behave in ways organisations can understand, manage, and trust over time.

(Photo by Roman)

See also: Autonomous AI systems depend on data governance

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.

AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.

The post As AI agents take on more tasks, governance becomes a priority appeared first on AI News.