NSA Gets 30-Day Access
The NSA now has pre-release access to powerful AI models, sparking concerns over surveillance and privacy
The National Security Agency has been granted 30-day access to powerful AI models before their public release, according to Executive Order 14409, signed by the President on June 2, 2026. This move has raised concerns among privacy advocates, who argue that such access could be used to compromise individual privacy and undermine data protection efforts. The NSA's new leads for key cybersecurity posts, announced on therecord.media, will likely play a crucial role in overseeing the agency's access to these AI models.
The recent breach of NSA classified systems by Mythos AI has highlighted the vulnerabilities of the agency's cybersecurity measures. Senator Mark Warner has called for increased efforts to strengthen these measures, citing the need to "put pedal to the metal" in the face of emerging threats. The incident has also raised questions about the effectiveness of the NSA's partnership with Anthropic, a company that has been patching dozens of security vulnerabilities in its flagship coding model, Claude Code. The loss of access to Anthropic's tool, as reported by nytimes.com, has significant implications for the agency's cybersecurity capabilities.
The 30-day access granted to the NSA has sparked concerns about the potential for abuse, particularly in light of recent developments such as the EUDI Wallet biometrics controversy. As the agency continues to expand its reach into the realm of AI and cybersecurity, it remains to be seen how these efforts will impact individual privacy and national security. The NSA's classified AI benchmarking for military "Frontier Models" is a significant development, and one that will likely be closely watched by privacy advocates and cybersecurity experts in the coming months.
Cross-reference independently — do not take our word for it.
Disclosure: NewsAnarchist uses AI-assisted reporting with web search. Always verify primary sources linked above.
This article is produced by NewsAnarchist's AI reporting system, not a human staff reporter. It's built from the primary source cited above (a FOIA release, an agency's own policy or procurement document, court filings from surveillance litigation, or the wire reporting linked in the body) and reports what that source states, attributed to it — it does not allege intent behind a surveillance program beyond what the record shows. Part of our Surveillance State hub. Found an error? Tell us.