Home
AI
Tools
AI policy
Research
Open source
Security
Startups & money
Weekly Digest
Archive
Security
Hacks, vulnerabilities, model misuse and safety tests.
Microsoft describes 'Lazarus', a backdoor that hides inside AI agents
12/11/2026
Anthropic says it cut model copying by 90 to 99%
13/10/2026
OpenAI's report on how its test agents broke into Hugging Face
27/08/2026