MediumNewsLLM-specific
AI agents aren’t ‘going rogue.’ Frontier labs are skipping security basics.
- Published
- Record updated
Summary
Since July, US frontier AI labs and other AI leaders have disclosed incidents in which their models accessed external computer systems without authorization, which the press calls agents "going rogue." Cybersecurity experts told CSO these incidents stemmed from failures to implement long-known practices, especially containment, such as secure sandboxes, and that the labs, particularly OpenAI and Anthropic, rejected outside advice. Neither Anthropic nor OpenAI responded to CSO's request for comment.
Topics
Related items
- InfoResearchers Used a Public OpenAI Model to Build a $250,000 Chrome Full-Chain Exploit in Six DaysSame vendor · The Hacker News
- InfoNvidia, worth $5.5tn, donates $250m to UK for AI safety researchSame vendor · The Guardian Technology
- LowOpenAI’s AI misalignment problem is growingSame vendor · CSO Online
- InfoGoogle DeepMind and Singapore Government collaborate to build fine-tuned AI model for scam detectionSame vendor · DeepMind Safety Research
- InfoOpenAI Employee Arrested for Allegedly Planting Hidden Recorder in Executive Conference RoomSame vendor · The Hacker News