Anthropic, the creator of the Claude chatbot, wrote to Governor Gavin Newsom urging the state’s courts to require kill‑switch mechanisms, external safety audits, and other regulatory safeguards for high‑risk artificial‑intelligence models.
Anthropic, the maker of the Claude AI chatbot, has formally appealed to California’s judicial system, urging judges to mandate robust safety controls for advanced artificial‑intelligence models.
Letter to Governor Newsom
In a detailed letter addressed to Governor Gavin Newsom, Anthropic’s leadership outlined a series of safeguards it believes should be legally required for high‑risk AI systems. The proposals include mandatory kill‑switch mechanisms, independent external safety audits, and continuous oversight by state‑appointed experts.
Proposed Safety Measures
- Kill‑switch capability that can immediately halt model operation in emergencies
- Regular third‑party safety audits to assess alignment and risk
- Transparent reporting of model capabilities and limitations to regulators
- Ongoing monitoring by a state‑designated AI safety board
Anthropic argues that these measures would help prevent unintended harms, such as disinformation amplification or autonomous decision‑making that conflicts with public policy.
Industry Context
The request arrives amid growing calls from tech firms, civil‑rights groups, and lawmakers for clearer AI governance. While some companies favor self‑regulation, Anthropic’s stance reflects a shift toward seeking judicial backing for enforceable standards.
Critics note that imposing legal requirements could slow innovation, yet Anthropic maintains that responsible development must precede rapid deployment.
For readers interested in broader perspectives, see Opinion section on uxc.news and the company’s profile Anthropic on Momentum.