Anthropic opens Singapore office, signs AI safety agreement with government

Anthropic opens Singapore office, signs AI safety agreement with government The agreement covers AI safety, model evaluation, responsible adoption and incident response. SINGAPORE: Singapore and artificial intelligence (AI) company Anthropic have inked an agreement on AI cooperation, as the firm behind frontier model Claude opened its Singapore office on Wednesday (Oct 14). Among other areas, the agreement sets out how Anthropic can work with the government on matters concerning safety and security, as well as the adoption of the tech and its impact. Speaking at the Anthropic Builder Summit at Capella Singapore on Wednesday, the firm’s co-founder Benjamin Mann said the memorandum of understanding (MOU) lays the foundation for cooperation between the frontier AI firm and the Singapore government, with safety evaluation among the most "exciting" areas. "When we train a new model, nobody really knows exactly what its capabilities are. Working with AI safety institutes like the new one in Singapore, as well as those around the world, is very important to understanding the risks and benefits of these new models as they’re being developed,” he said, referring to the Singapore AI Safety Institute. He added that the MOU can deepen understanding of public-private partnerships to ensure companies develop this new technology safely and responsibly, and to understand the technology's impacts on the world and the workforce. The MOU was signed on the sidelines of the Singapore International Cyber Week (SICW) conference on Tuesday by Anthropic co-founder BenjaminMann and Mr Chng Kai Fong, permanent secretary of the Ministry of Digital Development and Information (MDDI). Minister for Digital Development and Information Josephine Teo and Anthropic head of APAC policy Michael Sellitto witnessed the signing. The deal comes after Senior Minister and Coordinating Minister for National Security K Shanmugam said in an interview on Tuesday that the government hopes for “full transparency immediately” and quicker sharing of information amid rising risks from AI. “The key ask from Singapore, as well as many of the other countries, was: when some things happen, can we have full transparency immediately?" said Mr Shanmugam, referring to the major tech firms that have opened offices in Singapore. Apart from Anthropic, Singapore has signed several agreements earlier this year with major AI firms. In May, the country inked a deal with OpenAI – which is behind ChatGPT – to strengthen its position as a leading hub for applied, AI-native innovation. In June, the Infocomm Media Development Authority (IMDA) and Microsoft signed an MOU to deepen collaboration on AI safety and security, as both sides seek to address growing risks from rapidly advancing AI technologies. Then, as part of the government's smart nation strategy and cybersecurity efforts, the Government Technology Agency of Singapore (GovTech) is collaborating with artificial intelligence research laboratory Google DeepMind to develop real-time detection of scams on devices. NEW OFFICE AT RAFFLES PLACE As for the new Singapore office, which is located at Raffles Place, Anthropic remained coy about the Singapore team's headcount or whether it is looking to hire more people. There are currently 12 job openings listed on Anthropic's website, with roles ranging from finance to sales. Said Anthropic’s ASEAN general manager Dale Finlay: "The big focus for us now is very much on the ‘go-to-market’. We've got an amazing customer base here already across public sector, enterprise, digital native, startup, and so we're going to be building that team out, but we're not sharing numbers on this.” The company said its work with Singapore goes back to 2023 when it joined an evaluation sandbox run by IMDA and the AI Verify Foundation. Mr Mann said Anthropic worked with the Singaporean government three years ago, but the company could not do much at the time because the models were not capable enough. "In the beginning, our models were very bad," said Mr Mann in a fireside chat moderated by Mr Finlay. "Yet, we continued to make the case that the models would get better, that they could do things like increase state capacity and better the lives of all the citizens in Singapore and elsewhere, and eventually that vision came true." Mr Sellitto, head of APAC policy, said one reason Anthropic is excited to work with Singapore is that the government is very "forward-leaning" about the responsible adoption, use, and regulation of technology. "Actually, the government of Singapore was one of the first foreign governments that came to visit Anthropic in the very early days of the company," he said. "They saw where the technology was going and really had the foresight to realise how important it was going to be to the economy and society." HOW SINGAPORE USES CLAUDE The firm’s co-founder Mr Mann said it is “inevitable” that there will be large job displacements caused by AI, though Anthropic has yet to see that happening. "We think that this will happen, and our efforts on the Anthropic Economic Index help both us and governments to understand how these things are happening," said Mr Mann. The Anthropic Economic Index is a public research initiative by Anthropic that analyses real-world data from anonymised Claude conversations to measure how artificial intelligence is being used across jobs, industries, and daily tasks. Its most recent index ranked Singapore second out of 121 countries in Claude.ai usage relative to population. Usage is 5.81 times higher than would be expected based on the size of its population. The company said in a statement that more than 1,500 GovTech engineers and technical staff have adopted Claude Code – Anthropic’s agentic coding tool – to modernise government systems and speed up software delivery. “This has helped GovTech Singapore teams build, maintain and enhance digital services more efficiently at national scale,” the firm said. In the private sector, Anthropic noted the use of Claude by various financial institutions, including sovereign wealth fund GIC. GIC is a major investor in Anthropic, participating in several rounds of fundraising for the AI firm amid talk of a highly anticipated initial public offering sometime this year. Said Mr Mann: "I think because we sort of had this consistent narrative throughout the years, that's led a lot of governments to trust us. "Keeping your eye on that long-term mission, even when it's difficult, will make things easier later on. Because it might be hard in the beginning, but then it'll be easier. Whereas the flip side, as you get bigger, if you don't have that mission, then it's hard to remain differentiated."