Japan asks Anthropic and OpenAI for more access to advanced AI models, Nikkei reports

• Government plans to include closer cooperation with US AI developers in a cybersecurity package that could be released as soon as next week.

Governments seeking preferential access to frontier AI models under a national security or cybersecurity banner is now an established pattern, and such requests have historically resolved into structured partnerships, sovereign licensing arrangements, or government-grade deployments rather than outright grants of access. The distinction that matters for the named US developers is between a commercial government contract, which is a revenue line, and a regulatory or export-control concession embedded in a security package, which touches the terms on which models are supplied across borders. The cybersecurity framing suggests the ask runs through defensive and resilience use cases, a channel that has tended to be easier to clear than military or offensive applications. For the US firms, prior form in comparable episodes has been to engage while seeking reciprocal commitments on market access and data handling, turning such requests into negotiated frameworks rather than unilateral disclosures. What follows in the usual sequence is the package text itself, any conditionality attached to access, and whether other allied governments make parallel requests, which is how a single bilateral ask becomes a template.