Meta Platforms CEO Mark Zuckerberg said on Tuesday that AI companies already have enough reason to build their systems safely on their own.https://live.euronext.com/en/financial-news/metas-zuckerberg-says-ai-labs-have-enough-incentive-build-safely
Zuckerberg said every AI lab has both the responsibility and the incentive to train its models carefully. He added that each company has the tools to make sure that happens without outside coordination.
His statement came three days after Anthropic CEO Dario Amodei asked AI firms to slow down how fast they improve their models. Amodei's request followed warnings from researchers about long-term risks tied to advanced AI systems.
OpenAI chief Sam Altman and xAI's Elon Musk both backed Amodei's call within hours of it being made. Their support pointed to some agreement among AI leaders on the need for caution.
Zuckerberg Breaks From the Slowdown Call
Zuckerberg did not join that call. Instead, he argued that trust and alignment are becoming the traits that separate strong AI models from weaker ones.
"Any lab that doesn't focus on alignment will fall behind," Zuckerberg wrote. He framed safety as a business advantage rather than a shared industry pledge.
He also pointed to potential legal exposure as a reason labs act carefully. Zuckerberg said companies would face heavy liability if their models caused harm.
As an example, he cited Meta's decision earlier this year to delay its Muse AI agent so the company could strengthen its security. He said Meta made that choice on its own, without anyone asking it to.
Zuckerberg said Meta has already put most of its computing power toward building products for everyday user needs. He said this differs from chasing systems that can improve themselves without human input.
He also said Meta Superintelligence Labs, the company's AI division, already uses independent evaluators to review its work. He called that approach an industry best practice that other labs could adopt.
Political and Regulatory Backdrop
President Trump addressed AI safety concerns on Monday, a day before Zuckerberg's post. He said the United States already has enough guardrails in place for AI.
Trump also said China would benefit if doubt was cast on American AI development. Nvidia CEO Jensen Huang has voiced a similar view, saying regulation could hurt U.S. competitiveness against Chinese firms.
Zuckerberg's comments follow Meta's agreement to pay up to $18 billion to settle U.S. state lawsuits. Those lawsuits alleged Meta designed Facebook and Instagram to be addictive to children. Meta did not admit wrongdoing in the settlement.
FTC Chairman Andrew Ferguson added another angle to the debate on Tuesday. He said people should be "deeply suspicious" of AI companies that seek antitrust exemptions while also asking for new regulations.
Ferguson did not name Anthropic directly in his comments. Last week, Amodei had asked for antitrust permission that would let rival companies coordinate a slower pace of AI development over safety concerns.
