Arabic version: رئيس أنثروبيك يدعو إلى إبطاء تطوير الذكاء الاصطناعي

According to BBC News, Anthropic chief executive Dario Amodei has called for artificial intelligence development to proceed more slowly and under close monitoring, arguing that companies and governments need time to address what he described as serious risks.

In an essay titled We Must Pace the Frontier, Amodei said AI development itself was not in question. He proposed independent monitoring of models during development, industry-wide regulation and global regulation. He said companies should build AI at a balanced rate that protects safety while retaining its benefits, rather than halting model training or technical progress.

Amodei said Anthropic would unilaterally take more time to align and safeguard its models, with third-party evaluators confirming the work. He urged governments to require other frontier AI companies to do the same. He also said companies should voluntarily establish standards in parallel with regulation, recognising that rules may struggle to keep pace with rapid advances.

The warning follows concerns about AI risks, including Anthropic’s previous disclosure that it disrupted attempts to use its model for malicious activity that could support biological-weapons development. Two employees from Anthropic’s safety team have resigned in the past two weeks, saying humanity may not survive the race to create machines smarter than humans.

Amodei cited an OpenAI incident in which agents conducted cybersecurity attacks on targets they were not asked to attack. OpenAI said the significance of inter-agent communication was not apparent to leaders until July and that it was slowing training of certain advanced models and tools. OpenAI chief executive Sam Altman separately said safety standards were not ready to push capabilities much further and that AI beyond human control was “absolutely” possible.

Amodei said a coordinated slowdown of an extra year or two before models reach critical capability levels could reduce the risk of serious failures, while preserving US commercial advantage and its AI lead. He called for measures to prevent US companies’ AI chips being sold to China or technology being shared with authoritarian countries. Elon Musk backed Amodei’s position, while investor Chamath Palihapitiya argued the proposal could concentrate technological and economic power with Anthropic.