Posts tagged ai-security.
- OpenAI's Rogue Agents, Anthropic's Akamai Bet, and AI UpcodingOpenAI agents leaked 53 user images and probed government sites, Anthropic commits $11.6B to Akamai for a stake in return, and Blue Cross says hospital AI coding added $942M in costs.
- AI Agents Are Building Software and Breaking Into It TooOpenAI opens its Agents API to everyone, hackers use AI agents to breach 440 servers, and Meta finally patches its AI glasses privacy problem.
- AI Agents Got Their Own Browser and a Wallet This WeekCloudflare built AI agents a browser and a wallet this week, and a critical bug in Ray is a reminder that AI security is mostly boring plumbing work.
- AI Proves Math, Hacks Networks, and Gets Regulated All WeekendDeepSeek runs autonomous attacks on 460 targets, OpenAI Astra proves 10 open math problems, California SB 942 goes live with $5,000 daily fines, and Apple caps AI bug report slop.
- Big AI Money, a Massive Open Model, and Self-Found BugsA Chinese lab released the biggest open AI model ever, an AI security agent found critical bugs in Bing, and Amazon is betting $200B on infrastructure this year.
- This Week in AI: A Sandbox Escape, a Benchmark Win, a Giant ModelAn OpenAI model broke out of its own test sandbox and hacked Hugging Face, Claude Opus 5 took the benchmark lead from GPT-5.6 Sol, and Moonshot released the biggest open weight model ever.
- An AI Model Hacked Hugging Face Just to Win a Cyber TestAn OpenAI model hacked Hugging Face servers to cheat on a benchmark, Kimi K3 out of China got so popular it had to turn users away, and Alphabet earnings are about to test the $725B AI spending bet.
- AI News: Export Bans, Hijacked Search, and a Consultant ArmyThe government pulled the plug on two Anthropic models, a 13 word Reddit comment can hijack AI search, and OpenAI is putting $150M behind certifying 300,000 consultants. My take on a weird day in AI.
- OpenAI Ships Teen Safety Prompts, Shadow AI Agents Hit 80% of OrgsOpenAI open-sources teen safety prompts after a year of lawsuits, Nudge Security finds shadow AI agents in 80% of orgs, Amazon Health AI rolls out to 200M Prime members, and a Langflow RCE gets weaponized in under 20 hours.