Anthropic Opens Singapore Office and Inks Safety Deal
Anthropic opens a Singapore office and signs an AI safety pact with the digital ministry, covering model audits while excluding access to unreleased tech.
The partnership with Singapore’s digital ministry covers model safety evaluations and civil servant training, while excluding early access to unreleased AI models.
Anthropic opened an office in Singapore’s Raffles Place financial hub on Wednesday (October 14) and signed a Memorandum of Understanding with the Ministry of Digital Development and Information (MDDI). Announced during Singapore International Cyber Week, the move formalizes the startup’s ties with local regulators and anchors its regional safety evaluation work in the city-state.
Technical evaluations and stress testing
The agreement, signed by Anthropic co-founder Ben Mann and MDDI Permanent Secretary Chng Kai Fong, focuses on direct collaboration with the Singapore AI Safety Institute. Technical teams will work together to evaluate frontier models before and after commercial deployment, develop rapid response protocols for cyber incidents, and exchange research publications, including data from the Anthropic Economic Index.
This institutional relationship predates the new physical address. Since 2023, the Claude maker has taken part in the Generative AI Evaluation Sandbox run by the Infocomm Media Development Authority (IMDA). Those trials included red-teaming exercises across four official regional languages: English, Mandarin, Malay, and Tamil. The new agreement also introduces training programs for public sector staff using AI tools and practical guidance for local small and medium-sized businesses.
The Anthropic Singapore office and enterprise ties
The Raffles Place location is Anthropic’s fifth operational hub across the Asia-Pacific, joining existing sites in Bengaluru, Seoul, Sydney, and Tokyo. The Singapore team reports to Dale Finlay, head of ASEAN operations, supporting corporate and public sector accounts that already deploy Claude in production. That roster includes DBS Bank, transport and payments company Grab, tech conglomerate Sea Group, and government agency GovTech Singapore.
Singapore has become a favored operational hub for AI developers because it created clear governance rules alongside technical evaluation sandboxes. For frontier model builders, operating directly in the country serves as a credible technical benchmark that carries weight with corporate and public buyers throughout Southeast Asia.
Scope boundaries and undisclosed financial terms
Despite the public signing, the partnership has explicit limits. Michael Sellitto, Anthropic’s head of APAC policy, stated that Singaporean agencies will not receive early access to unreleased commercial models. Joint testing remains restricted to recent publicly accessible systems and standard cyber verification protocols, ruling out priority testing on upcoming flagship releases.
Neither Anthropic nor the Singaporean government disclosed budget details or financial commitments under the memorandum. The company also declined to release hiring targets or headcount figures for the Raffles Place space. The collaboration centers on shared evaluation methods and local regulatory alignment, leaving exact investments and expansion figures private.