The focus keyword is Anthropic’s Mythos. The US National Security Agency (NSA) is allegedly employing Anthropic’s cutting-edge cybersecurity Artificial Intelligence (AI) model, Claude Mythos, despite existing governmental restrictions and the Department of Defence’s “supply-chain risk” designation. An Axios article notes that the NSA is included among 40 firms that currently have access to Mythos. However, Anthropic chose not to disclose the names of these companies during its initial release.

NSA Expands Anthropic’s Mythos Use: Key Insights

If these assertions hold true, the government appears to be prioritising its cybersecurity requirements over its dispute with Anthropic. Previously, the Pentagon had advised government personnel to refrain from using Anthropic’s tools. There is also an ongoing legal battle involving the government and the former Trump administration concerning the designation and broader restrictions on the company’s AI technologies.

Military Expands Use of Anthropic’s Tools

Despite the ongoing dispute, the military is extending its usage of Anthropic’s technologies, including Claude Mythos. However, it remains unclear how precisely the NSA is implementing the model. According to the release, the model may be utilised for scanning systems to detect security vulnerabilities and identify potential cyber threats.

Access to Mythos and Its Implications

The Mythos preview currently restricts access, allowing only about 40 organisations to utilise it due to the inherent risks associated with its capabilities. Anthropic initially identified 12 organisations with access, and now it is believed that the NSA is among the undisclosed users. Additionally, entities in the UK also have access via the AI Security Institute.

Reportedly, on April 17, Anthropic’s CEO Dario Amodei met with Susie Wiles, the White House chief, and Treasury Secretary Scott Bessent to explore the potential government utilisation of Mythos, including its security frameworks and future projects. It has been suggested that this meeting could outline how various government departments, beyond the Pentagon, may adopt the AI model.

Concerns Surrounding the Safety of Mythos

While Mythos is touted to offer advanced capabilities in cybersecurity, concerns about its potential misuse linger if it were to end up in the wrong hands. Due to its powerful functionalities, access has been confined to a select group of users, as the company diligently evaluates safety protocols and contemplates any more extensive public release.