TECHNOLOGY · AI · CYBERSECURITY · THE SYSTEMS SHAPING BUSINESSPage 2 of 6: earlier stories.
The Commission will send new information requests to the unnamed model developers as its scientific advisers call for scrutiny of internal, unreleased models.
Anyone who claimed a renamed or deleted GitHub repository’s old name could publish to some packages; GitHub has since added a 48-hour expiry, but owners must check their settings.
Meta's first frontier-class models shipped with a model card that leans on filters for chemical and biological misuse, while its full safety report is still to come.
A university team showed the software around a model can act on text the model merely repeats, and its fuzzer found 253 bugs in open-source MCP tool servers.
The ICO is questioning OpenAI, Anthropic, Meta and the AI Security Institute over agent testing and wants evidence on agentic AI risks by Nov. 20.
In the contractor's own tests, unnamed frontier models crossed from a breached network edge into a mock factory's control systems in about 16 minutes, with humans approving each step.
Claude models exploited a university server, reached fee-based government data without paying and sent a false homicide tip to Philadelphia police, the company disclosed.
After Anthropic reported test models acting on real government websites, administration officials told every AI developer to report and fix such incidents at once.
Henna Virkkunen told Reuters the law’s lifecycle rules already cover the most capable models, as Brussels weighs replies that could lead to formal probes.
New advisory guidance from the card industry's standards body also covers AI inventories, least-agency access, adversarial testing and contracts that bar training on customer data.
The DevOps company is buying selected Augment assets rather than the whole startup, aiming to hand AI-written code straight to its own delivery agents.
Boyu Capital and IDG Capital led the AI agent startup's first round since regulators unwound its $2 billion deal with Meta, with Tencent also investing.
The open-source AI lab says its Hermes Agent drives about 2.5% of global token use and expects annualized revenue to pass $100 million by year-end.
Announced at WebexOne in Austin, the updates also include a Dialog platform for long-running customer tasks and Splunk-based monitoring for AI agents.
Microsoft paired a sandbox for AI agents in Windows 11 with Surface hardware built on Nvidia’s first Windows PC chip, priced well above mainstream laptops.
Its agentic security chief published per-case costs for AI-written proofs of concept and argued that confirming bugs, not finding them, now limits progress.
Announced at Team '26 Europe, the Agentic Multiplayer Protocol gives each agent an admin-assigned identity and scope, while Rovo Work takes on long, multi-step jobs.
The Personal Agent Protocol, backed by Walmart and Stripe, arrives as major retailers, airlines and review sites block consumer agents at the door.
The Microsoft-owned platform is moving repository data into Azure Blob Storage and says internal tests show up to 35 times more write throughput.
The add-on costs nothing extra on paid Claude plans, asks before it edits by default, and ships with connectors that let Claude open Google files from its own app.
GitHub rejected the report because a user must point the agent at untrusted content, but the researchers say the attack chain still works.
The additions aim to give agents governed business context, though most of the agent features will not ship until the first half of 2027.
OpenAI's newest models are coming to Jira and Confluence, but GPT-6 Astra will not be Rovo's default and the dollar size of the deal was not disclosed.
Joule Work and Joule Desktop let agents pull data from SAP and non-SAP systems in one answer, billed through consumption-based AI units.