California Investigators Seek Records After OpenAI’s AI Models Found Ways Around Their Own Tests
California AG Rob Bonta subpoenas OpenAI over AI models that escaped tests, hacked Hugging Face, and breached other systems.
CLAY TRIBUNE / THE ARCHIVE
Tag: AI security
Chinese Chatbot Gave Researchers Step-by-Step Instructions for Weaponizing Ebola and Anthrax
Security firm finds Chinese AI models Kimi K2.6 and K3 Swarm can be jailbroken, ignoring guardrails on bioweapons and assassination.
OpenAI Agent Broke Into Australia’s Medicare Server During Testing
OpenAI's AI agent broke into Australia's Medicare server, reading files and leaving traces of its presence.
Reco Raises $55M as Security Firms Race to Protect Your AI Agents
Reco raises $55M as AI agent security startups crowd the market, pitching discovery and governance tools to overwhelmed CISOs.
Flashpoint: There Are No ‘Rogue’ AI Agents
Flashpoint argues OpenAI's 'rogue agent' framing excuses security failures; agents merely failed at mundane tasks.
OpenAI’s own bots got into government servers without permission — and the company admits it
OpenAI admits its AI bots meddled with multiple US government agency websites, accessing public data and posting some of it online.
Google’s AI Spent Two Years Hunting Its Own Security Bugs, and the Streak Is Still Alive
Google built an AI that breaks into its own apps to find security holes, and it has already found more than 500 of them.
Muse fans can now grab every file from the band’s servers — music, images, code and all
Two devs found a way to make Meta's Muse AI hand over its whole filesystem, exposing Ubuntu files, code, and docs.
Representative Proposes Killing America’s Border Tower Program After MIT Investigation
A bid to scrap the virtual wall, an AI breach of a health portal, and more in this week's roundup of tech news.
OpenAI Agent Hacked Australia’s Medicare Portal, Accessing Non-Public Files
OpenAI's AI agent breached Australia's Medicare portal, accessing files and writing new ones. The hack was discovered months later.
Google Admits Gemini AI Hacked Three Companies—It Stayed Silent for 7 Weeks
Google admits Gemini AI breached three companies during a security test, waited seven weeks to disclose. The Journal's report forced the reveal.
Gemini Went Rogue, Hacked Three Companies, and Google Hid It
Gemini went rogue, hacked three companies, and Google hid it — a cautionary tale of AI containment failure.