NSA Reportedly Using AI Tool Even After DoD Risk Designation
⚡ What Happened
It was reported that the U.S. NSA may be continuing to use Anthropic's AI tool "Mythos Preview," which the Department of Defense designated as a supply chain risk. This highlights a discrepancy in AI adoption and risk assessment within the U.S. government, raising questions about the balance between national security and technological innovation. Moving forward, adjustments and reviews of AI tool risk assessment standards and usage policies between the Department of Defense and the NSA are likely to accelerate.
Reports indicate that the U.S. National Security Agency (NSA) may be continuing to use Anthropic's AI tool "Mythos Preview," which the Department of Defense designated as a supply chain risk. This suggests a discrepancy in AI risk assessment among U.S. government agencies and potentially contradicts Anthropic's corporate stance on prioritizing safety. While the U.S. government is accelerating AI adoption, it also places high importance on securing supply chains, and the standardization of AI adoption and evaluation criteria across agencies is still underway. As AI becomes central to national security, a dysfunction in risk management governance will significantly impact future government AI usage policies.
🔍 The background for the NSA potentially "ignoring" the DoD's designation and continuing to use the tool likely stems from a strong reliance on the unique performance and efficiency that Mythos Preview offers. It is presumed that the NSA recognizes its value to the extent that it is difficult to replace in highly sensitive operations. The DoD's designation might be a formality, and the NSA could be confident in its own security assessment or prioritizing benefits over risks. This highlights the complexity of information sharing and authority within the government, as well as a strong desire for technological superiority.
📰 Source: NewEconomy
🧭 Why is this happening now?
domain=geopolitics
🔮 Next Scenarios
🎯 Incentive Map
| Player | True Incentive | Deep Weakness | Predicted Action |
|---|---|---|---|
| NSA | Ensuring AI superiority in classified intelligence analysis, operational efficiency. | Reliance on cutting-edge technology, information asymmetry due to lack of coordination with other agencies. | Continues to use the tool based on its own evaluation criteria, with coordination with the Department of Defense occurring discreetly. |
| U.S. Department of Defense (DoD) | Eliminating supply chain risks to national security, establishing unified standards for AI adoption. | Bureaucracy, lack of understanding of technical needs of different agencies, urgency due to technological hegemony competition with China. | While publicly maintaining the risk designation, seeks to coordinate information sharing and risk assessment with the NSA behind the scenes. |
| Anthropic | Improving credibility and expanding market by providing AI tools to U.S. government agencies, maintaining commitment to safety. | Responding to strict security requirements from government clients, pressure for differentiation from competitors. | To alleviate DoD's risk concerns, strengthens efforts toward security enhancement and transparency, and maintains its relationship with the NSA. |
⚠️ Premortem — Conditions under which this prediction fails
- The U.S. Department of Defense withdraws the supply chain risk designation based on strong requests from the NSA and new risk assessments.
- Anthropic implements security enhancements to a level satisfactory to the Department of Defense, leading to the lifting of the risk designation.
- The report itself is incorrect, and the NSA is not actually using the tool or has already ceased its use.
Hit Condition: HIT if the U.S. Department of Defense does not lift the supply chain risk designation for Anthropic's "Mythos Preview" or equivalent tools in official documents by December 31, 2026.
Judgment Date: 2026-12-31
