Pentagon's Complex Relationship with AI Startup Anthropic Amid National Security Concerns
Despite being labeled a 'supply chain risk,' Anthropic's advanced AI model 'Mythos' is seen as a critical national security asset, with potential for a deal despite ongoing disputes.
L'essentiel
- The Department of War faces a complex situation with AI startup Anthropic, labeling it a supply chain risk while acknowledging its advanced AI model, Mythos, as a critical national security asset.
- Despite past disagreements, signs point towards a potential resolution and continued collaboration.
Résumé généré par IA
Pourquoi c'est important
The Department of War (DOD) is navigating a complex relationship with AI startup Anthropic. While Anthropic's advanced AI model, Mythos, is recognized for its cybersecurity capabilities and potential national security value, the DOD has also designated the company as a 'supply chain risk' due to past disagreements.
The Department of War is currently having a complicated relationship with the AI startup Anthropic. On one hand, it has been labelled as a “supply chain risk”, while on the other, its latest model is being called an asset with Pentagon Chief Technology Officer (CTO) Emil Michael calling Mythos a critical “national security moment”.
“I think the Mythos issue... is a separate national security moment where we have to make sure our networks are hardened up,” Michael told CNBC’s Squawk Box, while noting that the government must evaluate the model's high-end capabilities to protect American infrastructure. Mythos is an AI model by Anthropic that offers advanced cyber capabilities. Unlike standard chatbots, Mythos is specifically designed to find and patch cyber vulnerabilities.
Anthropic's public feud with Pentagon
The relationship between the Pentagon and Anthropic soured earlier this year after the two sides failed to agree on how the agency could safely use the company’s models. This led the DOD to officially designate Anthropic as a supply chain risk. Despite the official “risk” label, the US military is reportedly already putting Anthropic’s tech to work. The DOD has used the company's models to support military efforts in the war in Iran, and the National Security Agency (NSA) is reportedly already using Mythos.
“NSA and Commerce evaluates all frontier models... to see what the capabilities are at the edge,” Michael explained, defending the need to study even high-risk technology.
Anthropic may get a peace deal
According to the report, there are signs that the two sides may come to a conclusion. Earlier this month, Anthropic CEO Dario Amodei met with senior Trump administration officials at the White House. President Trump later told CNBC that a deal between the company and the DOD is “possible,” calling the firm “very smart”. Michael added that the Pentagon still wants “guardrails” for AI use, but noted that these terms are “negotiable”.
Pentagon signs contracts with seven AI companies
While Anthropic remains in legal and regulatory fight, the Department of War is moving ahead with other partners with the agency announcing deals with seven AI companies to deploy technology across classified networks for “lawful operational use”. These include Google, Microsoft, Amazon Web Services, OpenAI, Nvidia, SpaceX and Reflection.
À surveiller
Perspective IA — des possibilités, pas des certitudes
A formal agreement or 'peace deal' will be reached between the Department of War and Anthropic.
Probable · En quelques mois
The Pentagon will continue to explore and integrate advanced AI capabilities, including those from 'high-risk' companies, into its operations.
Très probable · En quelques mois
Further details regarding the specific applications of AI in military operations, particularly concerning cybersecurity, will be revealed.
Possible · En quelques mois
Questions ouvertes
- What specific 'guardrails' does the Pentagon want for AI use, and what are Anthropic's counter-proposals?
- What are the precise capabilities of the Mythos model that make it a critical national security asset?
- What are the details of the 'war in Iran' where Anthropic's tech is reportedly being used?
- What are the specific terms of the deals announced with the seven other AI companies?
