Articles · Topic
Agents & security
5 articles
-
Agents & security The First Exploit of the AI Agent Era For twenty years, only a handful of people in the world could write a working exploit for a memory bug. Last Wednesday that stopped being true. Two vulnerabilities in the PHP core sat in the code for more than a decade, then a language model chained them together in a week. And now the finished script is circulating on the internet. -
Claude Code Anthropic's head of AI safety testing: We don't have a plan yet for aligning superintelligence At the beginning of September, Jacob Coxon announced he was leaving Anthropic. The name probably won't mean anything to you: he did pre-training research on models, earlier at OpenAI, most recently at Anthropic. He accompanied his resignation with a thread on X that took off immediately. In it, he claims that both OpenAI and Anthropic are heading toward self-improving superintelligence in a way he doesn't consider responsible. -
Agents & security OpenClaw 2.0 Does Multiplayer. Watch Out for It! It's tempting. One agent, one machine, a whole team in a single thread, with work handed off along with its context. The new version of OpenClaw lures you in with support for multiple users in one installation, in one thread. Limits? There are some! -
OpenAI & Codex The Case of OpenAI's Rogue AI Agents Attacking Hugging Face A month ago, a widely reported case saw OpenAI's AI agents attack Hugging Face servers. Now it has a resolution. The company itself published a technical report, and an independent investigation has come out as well. The two stories diverge somewhat, and they are more alarming than they first appeared. -
Agents & security The Agent Deleted Someone Else's Booking: What AI Agents Do to a Sloppy SaaS An Australian user asked OpenClaw to book a gym class. The agent found a missing ownership check in the booking API and canceled a stranger's reservation. A breakdown of what happened technically, and what it means for anyone running a SaaS.