The Pentagon’s cyber warfare division has established a special task force to expedite the integration of advanced AI tools with potent hacking capabilities, according to Politico, citing sources.
The initiative will bring together representatives from Cyber Command and the NSA to explore how the Pentagon can safely implement leading models across all aspects of its operations.
This involves assessing the potential use of neural networks in “high-level” systems that contain some of the most classified information.
The initiative underscores the Pentagon’s concern over the sudden emergence of privately developed neural networks capable of identifying vulnerabilities in digital systems faster than the world’s best hackers.
Amid these concerns, the Trump administration is preparing an executive order on AI oversight in cybersecurity. The document may introduce a voluntary scheme for AI developers to collaborate with U.S. authorities.
According to media reports, companies will be asked to submit their developments to the government 90 days before public release. Another proposal involves early access for critical infrastructure operators, including banks.
Discussion intensified after the release of Mythos by Anthropic. In April, the company announced that a preview version of the model had identified thousands of high-severity vulnerabilities, including in major operating systems and browsers.
Instead of a public release, the firm launched Project Glasswing—an initiative involving AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, Linux Foundation, Microsoft, Nvidia, and Palo Alto Networks to test the tool in secure environments.
In April, it was revealed that the NSA is using Mythos, despite the startup’s conflict with the Pentagon. It is unknown how the agency employs the company’s AI, but others use it to scan their own environments for vulnerabilities.
Disagreements between Anthropic and the Pentagon arose from the firm’s reluctance to allow the use of AI models for mass surveillance of U.S. citizens and the development of autonomous weapons. Consequently, Trump ordered all federal agencies to completely cease using the firm’s technologies within six months.
Earlier in May, the U.S. Army convened leading contractors to integrate AI into weaponry.