AnalysisCybersecurityJuly 22, 2026
Thomas Ptacek: Open weights models from 2025 could hack networks
Security researcher Thomas Ptacek argues open weights models from 2025, with a pentest harness, could perform sandbox escapes and hack most networks, and that OpenAI's sandboxes are less secure than assumed.
Featured · Thomas Ptacek
1 source
Daily brief
Get tomorrow's AI brief in your inbox
More stories today
- Contour transforms 2D topographic maps into 3D terrain
- TechCrunch Disrupt 2026 to host AI Stage
- Lilian Weng returns to OpenAI to work on recursive self-improvement
- OpenAI launches Codex Visualize skill for interactive charts
- PAROL6: Open-source 3D printed robotic arm design released