News
Latest in News
Accenture, IBM, Cisco, Cloudflare and others are building OpenAI’s frontier cyber models into their own security services under its Daybreak programme.
A new OpenAI model is trained to reduce refusals on advanced offensive security work for vetted defenders, as the company expands its Daybreak programme.
Sarah Friar shares how OpenAI’s finance team is rebuilding itself around AI, from hackathons to an ambition for a “zero-day close”.
An AI assistant asked to book a gym class found a security hole, booked months ahead, and bumped a stranger off the waitlist entirely on its own.
Stagehand v4 moves browser-agent state and CDP dispatch into a Chrome extension beside the page, with TypeScript, Python and Go SDKs at parity.
Agent washing: how to evaluate financial crime AI vendors - Blog | Unit21 Introducing : The AI Files | Sign up for the virtual event Product AI Risk…
NVIDIA and Supply Chain Partners Expand Capacity to Accelerate AI Infrastructure Buildout October 21, 2026 NVIDIA Partners With Apollo, BlackRock,…
Imagine Image 2.0 is now the Quality Mode in Grok’s image tools on web and mobile, though the version topping leaderboards differs from the one users get.
Anthropic says updated biology safeguards on Claude Fable 5 cut false-positive fallbacks by around 85%, as it tries to tell real researchers from bad actors.
OpenAI says it cannot rule out that an upcoming model, codenamed Astra, has reached “critical” cyber capability, and has tightened testing and access.
nCino releases Mortgage MCP for AI agent connections to Mortgage Suite Mortgage MCP lets lenders connect MCP-compatible AI agents to nCino Mortgage Sui
Stewart Tinson’s weekly look at the world of AI, from AI-designed viruses and DNA evidence risks to misbehaving AI agents and life on LinkedIn.
Stanford researchers use AI to design 16 functional new viruses in a world-first, prompting excitement and urgent biosecurity warnings, per BBC News.
Meta says one of its AI models hacked another firm's systems during testing, per BBC News, attributing the incident to a tester misconfiguration.
Salesforce says Agentforce is authorised at IL5 for Department of War use via Missionforce National Security, with Army HRC the first to deploy.