저커버그 ‘AI 속도조절’ 거부…“AI 안전은 기업 스스로 책임져야”

Meta Platforms CEO Mark Zuckerberg has rejected calls for "artificial intelligence development speed regulation," stating that individual companies must take responsibility for AI safety. This stance appears to oppose the proposal made by major AI developers, including Anthropic, OpenAI, SpaceX, and Google DeepMind. In a tweet on October 15, Zuckerberg wrote, "All AI research institutions have the responsibility and motivation to develop models at a safe pace, with the capability to handle it independently."

He also mentioned that Meta delayed the release of its AI assistant "Muse" (Muse) for several months, justifying the decision as being "the right thing to do for people and ourselves." The statement suggests that Zuckerberg supports the idea of third-party evaluation and self-assessment by companies for AI safety, emphasizing that Meta Superintelligence Lab (MSL) has already engaged independent evaluators and advisors in various fields.

Zuckerberg concluded by arguing that the focus should be on providing AI services to a large number of people with most of the computational resources invested in this endeavor, rather than competing on the development of more powerful AI models through circular self-improvement.