Skip to content
Timeline

AI Safety Level

Anthropic's former AI risk classification system based on capability thresholds; abandoned in April 2026 in favour of autonomy-focused threat models.

1 events

Filters
#616 Apr

Capability-based threshold framework abandoned by Anthropic in favour of autonomy threat models

AI: Jobs, Power & Money: Anthropic drops ASL, expands Glasswing partners
← Back to AI Safety Level