Anthropic’s new artificial intelligence model called Mythos has become a major topic of concern around the world. Experts and government officials are worried because the AI has strong abilities to find and handle weaknesses in computer systems. This raises important questions about both protecting national security and managing new technology risks.
The Department of Defense Chief Technology Officer, Emil Michael spoke clearly about the issue at CNBC’s “Squawk Box” on Friday. He explained that Anthropic itself is still considered a supply chain risk and remains blacklisted by the Pentagon. However, Michael highlighted that Mythos is a “separate national security moment”.
He said, “I think the Mythos issue that’s being dealt with government-wide, not just at Department [of] War, is a separate national security moment where we have to make sure that our networks are hardened up, because that model has capabilities that are particular to finding cyber vulnerabilities and patching them.”
These comments came after a heated public clash between the Department of Defense and Anthropic. The Pentagon also announced agreements with seven other AI companies to use their technology on classified networks for lawful purposes.
In simple terms, Mythos is very good at spotting hidden weaknesses in computer systems, including complex ones that combine multiple small problems. This makes it useful for defense but also potentially dangerous if used by bad actors. The National Security Agency is now testing Mythos in a secure project called Project Aether.
Read More: Anthropic Sues Trump Administration Over Blacklisting & ‘Supply Chain Risk’ Designation
The goal is to use the AI to find “zero-day” vulnerabilities secret flaws no one knew about before especially in Microsoft systems like Windows and Azure. It can also run “red-teaming” exercises, which means simulating attacks to test and strengthen defenses.
The UK’s Artificial Intelligence Safety Institute reviewed Mythos and called it a clear improvement over older models in terms of cyber threat. The institute noted that Mythos successfully completed a 32-step simulated cyber attack on its own, something it had not seen before.
Officials said the model works well against weaker systems, though results may vary against heavily protected ones. They added that AI tools like this will only become more powerful in the future.
The Chief Executive of the UK’s National Cyber Security Centre, Richard Horne, spoke at the CyberUK conference in Glasgow on 22-23 April 2026. He explained that Mythos should push companies to update old technology.
He said, “Frontier AI is rapidly enabling discovery and exploitation of existing vulnerabilities at scale, illustrating how quickly it will expose where fundamentals of cyber security are still to be addressed”.
Not everyone sees it as a complete revolution. Some cybersecurity experts pointed out that cheaper AI models can find similar problems, and most real attacks still come from simple issues like weak passwords or forgotten software updates.
Read More: US Federal Departments Start Transitioning Away From Anthropic as Phase-Out Begins
About 40 large companies, including Google, JPMorgan, and Goldman Sachs, have received early access to Mythos through a program called Project Glasswing. The aim is to test how the AI can help protect systems and to share what they learn with others.
The concerns have reached top financial levels as US Treasury Secretary, Scott Bessent held a meeting with leaders of major banks, including Goldman Sachs and Citi on 8 April 2026.
In the UK, Mythos was added to important meetings involving the Treasury, Bank of England, Financial Conduct Authority and National Cyber Security Centre.
Government studies of worst-case cyber attacks show worrying results. Payments could stop working, people might not receive salaries and banking services could go down. This could create panic and even cause runs on banks.
Anthropic has added safety features, such as making sure the model does not keep copies of sensitive code it studies. Further, the clear statements from Department of Defense CTO Emil Michael shows that governments are treating Mythos seriously.
Officials want stronger defenses and better rules as AI grows more advanced. The situation is a reminder that powerful new technology brings both opportunities and real risks that need careful handling.
Muhammad Usman Hashmi is a researcher in International Relations, focusing on climate diplomacy, global governance, and political economy in the Global South. He has contributed to policy dialogues with the Foreign Policy Community of Indonesia and serves as a Senior Research Fellow at the International Council on Human Rights, Peace and Politics. He is also associated with Rethinking Economics Islamabad, contributing to research on development and sustainability.
