AISimon Willisonabout 2 hours ago
Quoting Thomas Ptacek
1 min read
Thomas Ptacek stated that an open weights model from 2025 with a pentest harness could perform sandbox escapes and hack most networks. The surprise stems from assuming OpenAI has stronger sandboxes. Ptacek noted this does not require a frontier model.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
