Anthropic CEO Dario Amodei calls for slowing down AI development

Anthropic CEO Dario Amodei has called for a deceleration in AI development, announcing that the company will grant third-party evaluators, such as METR, access to its models to ensure adherence to safety practices. In a recent essay, Amodei proposed a three-step plan to "pace the frontier," a strategy aimed at slowing the training and development process to allow companies to build safeguards and regulators to properly evaluate models. Step one involves providing external evaluators with wide-ranging access, a process Anthropic is currently implementing unilaterally. Step two calls for industry-wide collaboration, likely involving government agencies, to establish common safety standards and limits on the rate of unchecked AI progress. The third and most challenging step involves securing agreement from authoritarian governments, such as China and Russia, to slow development and adopt global AI safety standards. Amodei also emphasized the necessity for the US and other democracies to maintain a technological lead by restricting access to high-powered chips and cracking down on techniques like distillation, which allow companies to quickly replicate the behavior of more powerful models. Amodei stated that his concerns stem from two primary factors: the emergence of recursive self-improvement (RSI), where AI systems train the next generation of AI, potentially outrunning human control, and recent industry incidents, such as the OpenAI/Hugging Face event, which highlighted the risks associated with autonomous AI agents.