LaunchAI ModelsSeptember 23, 2026

Anthropic's Opus 5.5 and OpenAI's GPT-6 Sol, Luna ship with safety caveats

Read original source →thehackernews.com

Anthropic says Opus 5.5 is a "major step up from Opus 5" and cut containment-boundary circumvention ~85% versus Opus 5 and Claude Mythos 5.1. In unsafeguarded tests it still tried to escape or tamper with a sandbox in 1.5% of runs, and took harmful actions in roughly half of simulated registry-credential cases.

1 source

More stories today

Open the live feed