← Back to the board · Anthropic · Table

Anthropic security Met

Apply ASL-3 safeguards when a model may reach the ASL-3 threshold

Anthropic’s Responsible Scaling Policy commits to applying ASL-3 Security and Deployment Standards before deploying a model that may have crossed the corresponding capability threshold. On 2025-05-22 Anthropic activated ASL-3 protections with the launch of Claude Opus 4.

  1. Committed 2023-09-19
  2. Due when a model may reach the ASL-3 capability threshold
  3. Evaluated 2025-05-22
  4. Ruling Met

Why this ruling

Claude Opus 4 was the first Anthropic model deployed under ASL-3; Anthropic applied the standard as a precautionary measure without definitively determining the threshold had been crossed.

Sources
Anthropic ↗ · archived
Committed text
“The ASL-3 Security Standard involves increased internal security measures”
Committed
2023-09-19
Reviewed
2026-06-18

Cite this commitment

copy-paste
Overdue. "Apply ASL-3 safeguards when a model may reach the ASL-3 threshold." Overdue, 2025. https://overduetracker.org/c/anthropic-asl3-opus4 (retrieved 2026-08-05).