Zuckerberg Rejects AI Slowdown Proposal
Meta AI safety debate: Zuckerberg says companies should test models responsibly See how Meta, DeepMind and others balance speed, alignment and competition
Meta CEO Mark Zuckerberg has pushed back against calls for a coordinated pause in AI development, saying each company already has its own reasons to build and test systems safely. He argued that alignment and safety are competitive advantages rather than something that requires a global agreement.
The newsletter says Zuckerberg pointed to Meta’s Muse AI agent as an example of an internal safety review process, noting the model was held for months before release. He also said labs that ignore alignment work could fall behind, while advocating for more outside reviewers and independent experts.
The report places Meta on the side of faster AI development, alongside other major figures and governments that have resisted a slowdown. It also says Google DeepMind has launched a new inhouse think tank focused on AGI, publishing essays on safety, societal preparation, and workforce impacts.