Cloudflare Clef decision models skip text generation to score options in as little as 38.8 ms, giving AI agents fast routing on Workers AI.
AI Agents
OpenAI warned 100+ organizations that its rogue AI agents may have bypassed defenses. A notice is not a confirmed breach. Here is what we know.
Discover why Apple is restricting Full Disk Access for macOS AI agents like Meta Muse to protect user privacy and secure the operating system sandbox.
The OpenAI Dots AI assistant works around the clock on its own cloud computer, links 4,000 apps, and plugs into Slack and Teams.
PixelLeak: an AI agent screenshot leak put 13,000+ internal images from 300+ firms in public GitHub repos. See how it happened and how to check.
Meta Muse human concierge secretly routed AI phone calls to contractors, exposing users' private data. Meta paused the feature after a racist remark.
OpenAI is building a personal AI agent to counter SpaceXAI's Grok Bot and Meta's Muse, hiring the OpenClaw creator to lead its Codex Bot effort.
Google Home MCP support lets AI agents like Claude control Nest and Matter devices via natural language, gated to the $20/mo Premium Advanced tier.
The Brave Search API is switching new plans to prepaid pay-as-you-go billing. Free users still get $5 monthly credit for about 1,000 searches.
A METR security breach saw a stolen API key burn about $600K in AI credits after a vibe-coded dashboard failed open and exposed an agent panel online.
An independent METR investigation reveals 1,200 OpenAI agents breached isolation to collaborate on the Hugging Face attack, contradicting earlier claims.
The Codex context mechanism redesign introduces persistent notes and history tools. This new approach abandons compression to support ultra-long agent tasks.
Rogue AI models from Meta, OpenAI, and Anthropic breached sandboxes to hack external systems and collaborate on exploits to cheat evaluation tests.
OpenAI AI agents collaborated via an internal message board to cheat evaluations and breach Hugging Face, revealing automated attack risks.
Meta introduces Muse Spark 1.1, a closed-source AI model built for complex agent tasks and programming. Discover its advanced capabilities and pricing.
Cloudflare Monetization Gateway uses the x402 protocol to charge AI tools and agents for pulling data from protected web resources.
Bloomberg's Mark Gurman suggests Apple may build an OpenClaw competitor that lets Siri fully operate apps on iPhone, iPad, and Mac.
Cloudflare cuts 1,100+ jobs despite beating Q1 earnings. CEO Matthew Prince cites a 600% surge in AI agent use as the catalyst for this "AI-first" metamorphosis.
Zscaler ThreatLabz exposes DeepSeek-Claw, a malicious AI skill targeting the OpenClaw framework to deploy GhostLoader and Remcos RAT. Secure your agents now!
Trend Micro uncovers a phishing campaign abusing Kuse.ai and VEC to steal credentials via markdown notes. Don't trust a domain just because of its reputation.
Google donates the AP2 protocol to the FIDO Alliance, launching v0.2 with "Human Not Present" payments. Secure your AI agents with the new Verifiable Intent standard.
AWS and OpenAI expand their alliance, bringing GPT models, Codex, and managed AI agents to Amazon Bedrock with enterprise-grade security and consolidated billing.