Anthropic has opened an office in Singapore, its fifth in the Asia-Pacific region, and signed a memorandum of understanding with the Singapore government. The company did not give a headcount or a dollar figure for either move.
The memorandum covers AI safety, security and model evaluation, research on adoption, and responsible adoption across the public and private sectors, according to the company. Co-founder Ben Mann signed it with Kai Fong Chng, permanent secretary at the digital ministry, on the sidelines of Singapore International Cyber Week. Minister Josephine Teo witnessed the signing.
The plans include technical exchanges with Singapore's safety and assurance institutions on model evaluation, testing standards and incident response. Anthropic also offered skills work and access to data from its Economic Index.
Rivals have moved faster on the ground. OpenAI opened its first applied AI lab outside the United States in Singapore in May, backed by S$300 million, about $235 million, and roughly 200 roles.
Anthropic's announcement focused on usage rather than investment. The company says more than 1,500 engineers at GovTech use Claude Code. Grab says its Claude-based sales assistant saves about 40,000 hours a quarter. DBS and GIC are rolling out the tool as well.
Where the company operates matters as much as the ceremony. Earlier reporting found that OpenAI and Google were selling advanced AI to Singapore-registered arms of Alibaba, Baidu and Tencent, whose parent companies appear on a Pentagon list of Chinese military-linked firms. US export controls name specific entities and places, and Singapore is not among the restricted locations.
Anthropic has taken a harder line on its own products. The company bars Chinese companies and their foreign-owned entities from using its frontier models, and it has pushed Washington to broaden export controls.
Europe is pursuing a similar kind of oversight, but through law. The EU's AI Office can demand documentation and request access to frontier models. It has struggled to get that access, including for Anthropic's Mythos model. Anthropic agreed to the EU's code of practice for general-purpose AI in July 2025. The company has also asked governments for the power to block deployments that pose a significant risk of catastrophic harm.
What outside evaluators are allowed to publish remains unsettled. Reuters reported Tuesday that Anthropic and OpenAI have sought control over what reviewers can make public, citing trade secrets. Anthropic says it retains sign-off only on new firms it evaluates.
Every figure in the Singapore announcement comes from Anthropic itself. That includes the claim that Singapore ranks second of 121 countries on the company's Economic Index. The index measures how much people use Claude, not how widely AI has been adopted across the economy.
The Singapore office is open and hiring, though the company has not set a target. On the same day, in Dublin, Anthropic's head of international said open models carry more risk by definition.
Cet article a été rédigé avec l'assistance de l'IA.
News Factory APP - actualités agentiques pour booster votre SEO et AEO.