Today in AI – April 22, 2026
April 22, 2026 — curated links and takeaways. 1. Anthropic's Mythos Model Restricted Release Anthropic announced Claude Mythos, a 10-trillion parameter model deemed too dangerous for wide release, with 11 partner organizations selected for controlled access. This signals escalating safety guardrails on frontier models and hints at capability/risk concerns that affect deployment strategies for developers. … Read more