AI’s leading executives are divided over what the industry should do to advance its most capable models. Anthropic’s CEO, Dario Amodei, proposed slowing the advances in AI so that safeguards can catch up. His proposal includes embedding independent evaluators in AI labs, with the cooperation and coordination of AI companies and governments. OpenAI’s CEO, Sam Altman, endorsed the idea of pacing the technology. Elon Musk agreed to some extent, particularly with the idea that companies should test each other’s models, but remains skeptical of more government oversight. NVIDIA’s CEO, Jensen Huang, and Meta’s CEO, Mark Zuckerberg both emphasized the responsibility of individual companies to develop safe products. The disagreement seems to turn on whether company-level accountability is sufficient or collective oversight will be necessary. As @downingARK wrote in this week's newsletter, we hope cool heads prevail, allowing the industry to address recent failures while continuing to deliver on the promise of AI’s benefits to humanity. Read the full newsletter ⬇️