NSA Using Anthropic Mythos Despite Official Blacklist
NSA deployed blacklisted Anthropic Mythos model per Axios documents cited by Reuters revealing inconsistent application of federal AI security rules.
Lede: Reuters reported April 19 2026 that the NSA is using Anthropic's Mythos AI model despite its placement on an official federal blacklist according to an Axios investigation (Reuters 2026; Axios 2026).
Axios obtained internal documents showing NSA deployment of Mythos for analytical workloads while the blacklist was created under National Security Council directives citing supply-chain and data-exfiltration risks (Axios 2026). A March 2024 GAO report documented similar uneven adoption patterns across DoD components after Executive Order 14110 established AI risk tiers (GAO-24-106723). NIST AI RMF 1.0 released in 2023 explicitly recommended uniform controls for high-impact use cases yet enforcement records remain fragmented (NIST 2023).
Original Axios and Reuters stories omitted direct references to parallel 2024 incidents in which CIA pilots bypassed identical restrictions on two other frontier models citing mission urgency per a redacted ODNI summary later obtained by The Washington Post (WaPo 2025).
AXIOM: Agencies continue to route around published AI blacklists when operational needs arise creating persistent gaps between written policy and actual deployment.
Sources (3)
- [1]Reuters: NSA is using Anthropic's Mythos despite blacklist(https://www.reuters.com/business/us-security-agency-is-using-anthropics-mythos-despite-blacklist-axios-reports-2026-04-19/)
- [2]GAO-24-106723 Federal AI Adoption Challenges(https://www.gao.gov/products/gao-24-106723)
- [3]NIST AI RMF 1.0(https://www.nist.gov/itl/ai-risk-management-framework)