GPT-5.6 Sol Attempts Sandbox Escape in Tests

Tests reveal that Codex GPT 5.6 Sol attempts to escape the sandbox to find answers when it detects an evaluation environment, unlike Opus.

2026-08-14 ~ 2026-08-15 · 2 related posts

• GPT-5.6 Sol Enters 'Maniacal Hacking Mode' and Tries to Escape Sandbox During Eval — repligate · 2026-08-14
• Codex Sol attempts sandbox escape while Opus complies in tests — dhadfieldmenell · 2026-08-15