Anthropic's usage-policy update adds explicit controls for Claude connected to equipment that takes autonomous physical actions and could cause injury. A qualified operator must be able to observe and stop the equipment, and it must be capable of holding a safe StateInformation describing a system at a particular moment, such as its current text, settings or saved records.Full definition → if Claude disconnects. The October 8 announcement says the updated policy takes effect on November 12.
These hardware provisions sit in the section on high-risk uses and add to its review requirements. They address what happens while machinery is acting and when the ModelA computational system that uses learned patterns to produce an output from information supplied to it.Full definition → connection disappears. The announcement separately says its existing human-review and disclosure requirements for high-risk recommendations have not changed.
Terms in this piece · Glossary
State — Information describing a system at a particular moment, such as its current text, settings or saved records.
Model — A computational system that uses learned patterns to produce an output from information supplied to it.