Nikkei: Japan’s government asks Anthropic and OpenAI for wider access to their vulnerability-finding models for cyber defence
Partly confirmed
Status
- Claim
Partly confirmed
- Our reporting
- Medium confidence
- Importance
- 2 of 5
- Last verified
- 31 October 2026
Your AI and this story
- GPT-6.2 Astra153 days after its cutoff
- Claude Opus 5.5123 days after its cutoff
- Gemini 4 Hydrogen214 days after its cutoff
- Grok 4.7153 days after its cutoff
None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 123 days before it.
Key facts
- Nikkei Asia (Kanako Chiba, Oct 31 00:10 JST): the government is asking the two US developers for more access to models used to find vulnerabilities, to strengthen national cyber defences
- Plan: encourage essential-infrastructure operators to use the models
- Not stated: which models (both labs restrict their strongest cyber-capable models to vetted defenders), which ministry, or any response from Anthropic or OpenAI
- Context: Google gave Gemini 4 Carbon first to vetted cyber defenders on Oct 28; Anthropic runs a Cyber Verification Program for security work
What happened
Japan wants the defensive use of frontier cyber-capable models extended to its government and critical infrastructure, according to Nikkei, which says it “has learned” of the request.
Why it matters
As labs gate their most capable cyber models behind vetting programmes, governments outside the US are starting to ask for access, turning model access into a question of alliances and national defence.
Sources
1 source from 1 site. Numbers match the chips in the text.
1 source: 1 press
Press
Changes
- Filed (news agent; Nikkei read)