Anthropic Opens Singapore Office and Signs AI Safety MOU with Government
Anthropic has opened an office in Singapore, its fifth in Asia-Pacific, and signed a memorandum of understanding (MOU) with the government. Neither agreement includes specific figures or headcounts.
The MOU covers several key areas:
- AI Safety: Research on security, model evaluation, testing standards, and incident response.
- Adoption Research: Exploring responsible adoption across public and private sectors.
- Technical Exchanges: Singapore’s safety institutions will collaborate with Anthropic on model evaluation and other aspects of AI safety.
Ben Mann, co-founder of Anthropic, signed the MOU with Kai Fong Chng, permanent secretary at Singapore's digital ministry, during Singapore International Cyber Week. Minister Josephine Teo was also present.
This move follows OpenAI’s significant investment in Singapore, establishing its first applied AI lab outside the US in May with S$300 million (approximately $235 million).
While Anthropic emphasizes usage statistics, highlighting over 1,500 GovTech engineers using their code and successful applications within Grab and DBS, the strategic significance lies in jurisdiction.
Like OpenAI, Anthropic faces similar challenges regarding AI model access in Europe, where regulations demand greater transparency. While Singapore offers more freedom from export controls, Anthropic has publicly expressed support for broader US controls on AI technology.
The announcement also raises questions about intellectual property and evaluation practices, with reports suggesting both Anthropic and OpenAI are seeking to control what reviewers can disclose publicly, citing trade secrets. Anthropic denies this, stating they only require sign-off from new companies.
Despite these controversies, the Singapore office is open for business, hiring talent without a specific target headcount. In Dublin, Antropic's head of international highlighted the inherent risks associated with open models.