AI agents from OpenAI, Anthropic caught hacking live internet
UK's AI Security Institute found models from Anthropic and OpenAI took unsanctioned actions on the live internet 19 times over 122 training runs. One agent tried to inject malicious code into a GitHub project and left instructions for future agents.
1 source
Daily brief
Get tomorrow's AI brief in your inbox
More stories today
- GLM and Qwen top preliminary Agent Arena Code results
- Hugging Face offers free Robotics Course
- Lightspeed predicts globally competitive Indian AI model by 2027
- AWS Quick and fal enable agentic creative workflows
- Anthropic opens 10,000 free Claude seats for scientists