All corrections
LessWrong February 24, 2026 at 07:46 PM

www.lesswrong.com/posts/JdKyWmzfv8EeBuxHC/exclusive-hegseth-gives-anthropic-unti...

2 corrections found

1
Claim
Anthropic's Claude is the only AI model currently used for the military's most sensitive work.
Correction

Public reporting shows the U.S. military/DoD is using other generative-AI systems on highly sensitive (including Top Secret/classified) networks, so Claude is not the only such model/tool in use.

Full reasoning

The post claims that Claude is the only AI model currently used for the military’s most sensitive work.

However, multiple credible sources report other generative-AI systems are being deployed/used on highly sensitive DoD networks:

  1. Scale AI / Donovan on Top Secret networks (reported Sept 17, 2025). The Register reports that Scale AI partnered with the Pentagon to deploy AI on Top Secret networks, and that the DoD would get access to Scale’s Donovan (described as “a generative AI”) as part of that deal. If generative AI like Donovan is being deployed on Top Secret networks, Claude cannot be the only AI model/tool used for the military’s most sensitive work.

  2. Grok integrated into DoD networks alongside Google’s AI systems (reported Jan 2026). The Associated Press reports that the DoD will integrate Elon Musk’s Grok AI chatbot into its network “alongside Google’s AI systems,” with increased access to both unclassified and classified military data. That also contradicts the idea that Claude is the only AI model currently used for the most sensitive work.

Because credible reporting indicates other AI systems are deployed/used in sensitive/classified contexts, the post’s “only” claim is not accurate.

3 sources
2
Claim
Claude, which is currently the only model used in classified systems.
Correction

Reporting indicates other AI systems are deployed/used on DoD Top Secret/classified networks (e.g., Scale AI’s Donovan; Grok alongside Google AI systems), contradicting the claim that Claude is the only model used in classified systems.

Full reasoning

This sentence makes a very specific exclusivity claim: that Claude is “currently the only model used in classified systems.”

That exclusivity is contradicted by credible reporting that other generative-AI systems are deployed/used in classified/Top Secret contexts:

  • Scale AI on Top Secret networks: The Register reports a DoD/Scale deal to deploy Scale’s AI on Top Secret networks, including access to its Donovan generative-AI platform.
  • Grok + Google AI systems with classified data: The Associated Press reports DoD will integrate Grok into its network alongside Google’s AI systems, with increased access to classified military data.

Given these reports, Claude cannot accurately be described as the only model used in classified systems as of the post’s publication date (Feb 24, 2026).

3 sources
Model: OPENAI_GPT_5 Prompt: v1.6.0