The US National Security Agency (NSA) has reportedly lost access to Anthropic's advanced AI model, Mythos 5. This model was used for detecting software vulnerabilities, according to The New York Times.
Reasons for Access Restriction
The loss of access occurred after the Trump administration previously imposed export restrictions on Anthropic earlier this month, citing national security concerns. According to The New York Times, this decision deprived the intelligence agency of a tool that analysts considered highly effective in identifying software weaknesses.
Use and Issues with the Technology
The Associated Press reported on Wednesday, citing an anonymous US official, that the AI model could detect vulnerabilities in highly secure government networks in just a few hours during testing. Anthropic's technologies are increasingly being used in secret US government networks for intelligence analysis, operational planning, and cyber operations.
In February, the US Department of Defense reportedly classified Anthropic as a 'supply chain risk' after the company refused to lift restrictions on certain military applications of its AI systems. Anthropic stated that it opposes using its technology for mass domestic surveillance and fully autonomous weapons. Subsequently, President Donald Trump ordered federal agencies to phase out Anthropic technologies over six months.
Legal Battles and Warnings
Anthropic is challenging these measures in court, arguing that they constitute an illegal response to the company's refusal to ease safeguards governing the military use of its AI models. Despite the decommissioning order and ongoing legal dispute, some media reports suggest that parts of the US government continue to use Anthropic's AI systems.
These events come against a backdrop of growing warnings from researchers, tech leaders, and security personnel that the integration of AI into military and intelligence operations is happening faster than governments can adapt to its capabilities. Experts caution that the same systems used to strengthen cyber defenses could automate cyberattacks and lower barriers for malicious actors.
Five Eyes Alliance Warning
On Monday, cybersecurity experts from the Five Eyes intelligence alliance—which includes the United States, the United Kingdom, Australia, Canada, and New Zealand—warned that increasingly sophisticated AI models could soon give hackers the ability to disrupt governments, businesses, and critical infrastructure globally.
