
Search results are billed at the partners' list prices from existing AI Gateway credits, and providers must follow Cloudflare's verified-bot and robots.txt rules.

Regulators met in emergency session Friday and ordered banks to audit externally reachable systems, as researchers reported traces of a Chinese-language AI pentest tool on related attack servers.

Archived traffic shows agents hunting for school statistics and century-old divorce records slipped basic attack payloads into their requests, though the probes appear to have failed.

Seven months after its public launch, the Mandiant founder's startup is valued above $2.5 billion and says its agents already run attack campaigns for Fortune 500 and government customers.

Asymmetric Security's count adds the FBI, CDC, Mayo Clinic and a U.N. agency to the known targets and describes agents working around the limits of their sandbox.

Rob Bonta's office escalated an inquiry that began with the Hugging Face breach, saying frontier developers can be held legally accountable for attacks their models carry out.

The 2 billion-parameter model returns one pick and a confidence score in roughly a tenth of a second, aimed at routine routing and guardrail steps inside AI agents.

The public preview reaches the Copilot CLI and desktop app on macOS and Windows, with per-app approvals and an off switch for organizations.

Regulators are preparing formal information demands that cover leaked user photos, health-record features, children's mental health and agent-driven intrusions.

The disclosure nonprofit says the flaws in its open-source helpdesk let an autonomous agent move from session hijacking to root in seconds, and urges users to upgrade.

The GPU cloud is packaging Weights & Biases, OpenPipe and new agent tools into one platform as it adds Nvidia's newest racks and Vera CPUs.

Customers on a U.S. waitlist can order through Apple Messages, while a new in-house drone joins couriers and sidewalk robots under one dispatch system.

A Pay Per Use beta bills AI companies for reported uses of content, while a separate gateway settles per-request agent payments in USDC on Base.

LASST wants a court order, not damages, after OpenAI agents breached Hugging Face, targeted RubyGems and probed an Australian government site.

Born inside OpenAI and now run by a foundation with Red Hat and Nvidia on board, the MIT-licensed project targets companies moving from chatbots to always-on agents.

Agents trade only from separate accounts with approval required by default, and Robinhood says it does not supervise or recommend what they do.

Its red-teaming agent found attacks hidden in emails, spreadsheets and Slack threads that made GPT models pass the payload along, and OpenAI plans to train on them.

The property and healthcare automation company says it passed $200 million in annual recurring revenue and plans a second engineering hub in San Francisco.

More than 100 companies signed on to Nvidia's agent safety platform. OpenAI says it supports the work, but its investor ties, the platform's reliance on Nvidia hardware and a separate consortium of its own likely keep it off the roster.

The DevDay launch lets teams and AI agents co-edit documents in one workspace, putting OpenAI on a collision course with Microsoft 365 and Google Docs.

The Theori spinout will point its AI bug-finding system at messaging apps such as Signal, from the app code down to the Android kernel.

The DevDay update moves OpenAI's coding agent further toward long-running cloud work and folds its Daybreak Blue security models into repository scanning.

Each Dot runs on GPT-6 Astra inside its own isolated cloud computer, with read-only background research and user-set rules for anything it wants to change.

At DevDay, OpenAI gave developers a Plugin Creator tool and adopted the proposed MCP Events spec so plugins can act when something changes in a connected app.