# Anthropic reportedly embeds engineers inside the NSA

Anthropic has reportedly sent forward deployed engineers to work inside the US National Security Agency, as the intelligence agency explores ways to use the company’s artificial intelligence tools in cyber operations.

The arrangement highlights how quickly large language models are moving into sensitive government environments, including national security and cybersecurity work. It also shows that AI vendors are increasingly positioning staff close to customers to help adapt their products for specialized uses.

According to the source material, the NSA has been using Anthropic’s technology, referred to as Mythos in the report, in connection with cyber attack activity. The details available do not clarify exactly how the system is being deployed or what role the engineers are playing day to day, but the placement suggests a deeper level of technical collaboration than a standard commercial software contract.

Forward deployed engineers are typically technical staff placed on or near a customer site to help integrate a product, troubleshoot problems and tailor features to operational needs. For an AI company, the practice can be especially useful when working with institutions that have complex security requirements and highly specific workflows.

The reported collaboration comes as governments consider both the benefits and risks of using advanced AI in security settings. On one hand, the tools can help analysts process large volumes of information more quickly, identify patterns and support defensive cyber work. On the other, the same systems raise questions about control, misuse and the possibility of assisting offensive operations.

The NSA is one of the most prominent intelligence agencies in the United States and has long been central to the country’s cyber capabilities. Any use of commercial AI systems in that environment is likely to attract scrutiny from policymakers, security experts and the public.

Anthropic has been one of the more prominent AI developers to market its systems to enterprise and government customers, alongside rivals that are also seeking a foothold in regulated industries. The company’s work with the NSA reflects a broader trend in which AI providers are building dedicated teams to support large institutions that want to deploy models securely and at scale.

The source did not indicate whether the partnership is temporary or part of a longer-term arrangement, nor did it provide details on the number of employees involved. It also did not specify whether other US intelligence or defense agencies are receiving similar support.

Still, the report points to a growing reality in the AI sector. As models become more capable, the most strategic uses may not be in consumer products, but in tightly controlled environments where technical expertise, security clearance and close customer support are all required.