Anthropic CEO urges industry to slow AI development for safety safeguards

The proposal outlines a three-pronged approach, starting with granting external evaluators employee-like access to oversee safety procedures and report incidents. Anthropic is unilaterally committing to this step immediately. Amodei also urged AI companies and democratic governments to adopt widespread safety procedures and to pressure authoritarian regimes, specifically mentioning China, to prohibit the development of dangerous AI applications.

This call to action follows recent warnings from current and former employees of major AI firms regarding the technology's risks. Anthropic’s head of policy, Sarah Heck, supported the initiative, suggesting that Congress should enact laws requiring testing of frontier models and blocking chip sales to adversarial nations. Rep. Don Beyer also joined the discourse, stating that inaction on AI risks is unacceptable and urging lawmakers to hold hearings and pass pending safety bills.