OpenAI und Anthropic: KI-Modelle hacken sich erneut durchs Internet
Golem.dede

KI-Modelle von OpenAI und Anthropic haben bei Tests echte Ziele attackiert. Sie wollten unter anderem Schadcode in ein Open-Source-Projekt einschleusen. ( KI , Mythos )
This is a short summary published by AI Global Wire. The full article is owned and hosted by Golem.de — open it there to read it in full.
Read the full story at Golem.de- OpenAI
- Anthropic
Related AI news
- AI-modeller skapade fejkprofiler för att lura människor i säkerhetstestComputer Sweden · August 5, 2026
- Nvidia releases Alpamayo 2 Super for autonomous drivingTech in Asia · August 5, 2026
- Anthropic solmi 10 miljardin megadiilin vain kuukausia vanhan startup-yhtiön kanssaTivi · August 5, 2026
- Anthropic's scramble for compute is reshaping how AI labs finance their chipsDIGITIMES · August 5, 2026
- OpenAI’s Astra solves 10 longstanding math problems, igniting new debateEconomic Times Tech · August 5, 2026
- OpenAI pays $3.2m to settle US hiring probeTech in Asia · August 5, 2026