Anthropic CEO outlines plan to ‘pace the frontier’
techcrunch ·
Anthropic CEO Dario Amodei published a blog post echoing calls to “pace the frontier” of AI development and outlined three broad strategies, saying Anthropic would unilaterally commit to one. He said two developments convinced him to take a more cautious approach: the OpenAI-HuggingFace hack and the rapid recent acceleration of AI, especially its growing ability to build the next generation of AI. His first proposal is to allow embedded evaluators from third-party organizations such as METR to verify that AI companies follow their safety and pacing commitments and to ensure incidents are reported; Anthropic is committing to this and wants governments to require other frontier companies to match. Second, he called for leading AI companies in democratic countries to coordinate common safety standards and limits on unchecked progress, acknowledging concerns about antitrust and suggesting the US government could mediate or issue a narrow waiver for safety conversations. Third, he called for global coordination with authoritarian governments, including cooperation with China, while admitting strict limits and suggesting at least narrow agreements such as prohibiting AI use for biological weapons. Amodei also argued that measures such as refusing to sell powerful chips or semiconductor manufacturing equipment to Chinese companies and cracking down on model distillation could slow China enough to widen America’s lead significantly over the next three to five years. The post followed researcher Jacob Coxon’s resignation from Anthropic over concerns that leading AI companies are gambling with lives, though Amodei did not explicitly mention it; critics have called Amodei a doomer or said such proposals amount to regulatory capture, while he said the backlash is a crisis of trust and that he still believes AI can greatly improve human life if built correctly.
techcrunch