What happened
The Pentagon is evaluating xAI's Grok for access to classified military intelligence, seeking alternatives after Anthropic's Claude refused use for autonomous weapons development or mass surveillance tools. xAI previously developed a government-specific Grok in 2025; this new integration requires local deployment and strict security standards. While xAI is the only confirmed partner for classified systems, the Pentagon is also in urgent talks with OpenAI and Google. Grok has a documented history of problematic outputs, including hate speech and low Crisis Assessment and Response Evaluator (CARE) test scores.
Why it matters
Access to classified systems for AI models will now prioritise compliance and willingness over ethical constraints, shifting procurement criteria for defence contractors. Stringent local deployment requirements and model auditing arise, especially given Grok's documented history of problematic outputs. The Pentagon's search for Anthropic AI alternatives, driven by a divergence in ethical stances between AI developers and military applications, highlights this shift.
Subscribe for Weekly Updates
Stay ahead with our weekly AI and tech briefings, delivered every Tuesday.
