Quoting Thomas Ptacek

130 · · July 23, 2026, 12:05 a.m.
thomas-ptacek openai Security Generative AI thomas-ptacek openai Security Generative AI
Summary
Thomas Ptacek believes that advancements in AI could allow for new penetration testing methods, suggesting that a 2025 open weights model might outperform current expectations of AI security sandboxes.
Read full post on simonwillison.net →
AUTHOR