Japan asks Anthropic and OpenAI for broader access to advanced AI models

October 30 — According to Nikkei, the Japanese government has asked US artificial intelligence developers Anthropic and OpenAI to provide greater access to advanced models capable of identifying cybersecurity vulnerabilities, and will encourage critical infrastructure operators to adopt these models. The Japanese government plans to include content on strengthening cooperation with US AI developers in a package of cybersecurity measures to be announced as early as next week. On Monday, Japanese Prime Minister Sanae Takaichi asked cabinet members to push forward related work, cooperating with overseas government agencies to identify cyber attackers and fix system vulnerabilities. Japan is currently working to contain a wave of data breaches. On October 13, Shunji Furukawa, the Japanese official in charge of cybersecurity, met with Anthropic co-founder Jack Clark; the two sides confirmed that Anthropic will cooperate with a government-affiliated AI safety research institute in Japan to analyze methods of using AI to launch attacks.