Claude’s new physical-action rules require independent equipment limits

130 words · 1 min

Anthropic’s revised Usage Policy takes effect November 12 following its October 8 announcement.

For high-risk physical actions, it requires qualified human oversight and operating limits enforced by equipment or a controller independent of the model’s output. [1][2]

The operator must be able to observe and stop the equipment. If the operator intervenes or the connection to Anthropic’s services is lost, equipment must stop or hold a safe state. A prompt instructing a model to stay within a reach limit cannot substitute for that separately enforced control. [1]

For an equipment project, identify the component that enforces reach, speed or force, the person able to intervene, and the state reached after communication loss. Those controls must match the actual machine and environment. The policy does not replace applicable machinery or workplace safety requirements. [1]

Sources & details 2 sources

Discuss in your voice app

Take the reporting, visual explanations and original sources into a conversation. Copy the brief, paste it into your preferred assistant, then start its voice mode.

Open ChatGPT

Paste the brief into your voice app to discuss this reporting.

Revised 12 November 2026. Revised visual explanations and copy; added Nvidia investigation context and clarified tool eligibility.

Reporting sources

Anthropic — Usage Policy
Effective November 12; high-risk physical-action controls and independent limits.
Anthropic — October 8 policy announcement
Announcement date and November 12 effective date; identifies added physical controls.

Images & figures

A model-requested target beyond a separately enforced reach boundary is blocked. If the operator intervenes or the service connection fails, equipment must stop or hold a safe state.
Euangelion · Source · Usage terms · Download figure