- Access Trump’s executive order regarding AI models here.
“Within 60 days of the date of this order…. shall develop and maintain a classified benchmarking process to assess the advanced cyber capabilities of AI models and determine the threshold at which an AI model should be designated a ‘covered frontier model’ for the purposes of this order,” read US President Donald Trump’s Executive Order issued on June 2, 2026. He directed this particular section to the Secretary of War, Homeland Security and Treasury, among others, including the Director of the National Security Agency.
The US government seeks ‘voluntary’ AI model access before the release for security and IP protection: The executive order also states the following actions directed at various US departments and institutions to design a “voluntary framework for AI developers” to “provide the federal government with access to covered frontier models.” These are subject to confidentiality, cybersecurity, and intellectual property protections for up to 30 days before releasing models to other trusted partners.
It further seeks their collaboration “to promote secure innovation and strengthen the cybersecurity of critical infrastructure.” However, it also clarified that this order doesn’t require licenses, approvals or permits before companies can create, publish, or release AI models.
Anthropic welcomes Trump’s order: “This Executive Order is an important step in strengthening America’s leadership in AI. We look forward to collaborating with the White House to support its implementation,” read Anthropic’s post on X.
Read more about the almost-settled legal conflict between Anthropic and the US government here: [ Link-1 | Link-2 | Link-3 ]
What is a frontier AI model? “Frontier models are the most advanced AI models available at a given moment, trained on massive datasets to deliver state-of-the-art performance across many tasks, representing the leading edge of AI capability,” according to NVIDIA’s blogpost.
What is the US government’s intent? US government said that it wants to collaborate with the “private sector to modernise government and private sector information systems” to:
- Protect American innovation and intellectual property from being stolen or exploited by hostile actors
- Build America’s advanced AI capabilities
What are the other directives? The executive order further directed various departments and authorities to comply with the following actions within a timeframe of 30-90 days.
Within 30 days:
- Committee on National Security Systems: Strengthen cyber defense of national security systems (definition).
- Secretary of War: Strengthen the cyber defense of the Department of War systems.
- Secretary of Homeland Security: Issue guidance to:
- Prioritize cyber defense of civilian federal systems
- Expand AI-powered cybersecurity tools and services
- Give federal agencies, state/local authorities, and critical infrastructure operators (hospitals, banks, utilities) access to cybersecurity tools and frontier AI models
- Secretary of the Treasury: Create an AI cybersecurity clearinghouse with industry and infrastructure operators to coordinate vulnerability scanning, discovery, and patch distribution.
- Director of OMB: Identify federal grant programs that can fund AI vulnerability detection tools.
Within 60 days: the Director of the Office of Personnel Management is required to expand hiring for cybersecurity specialists in the US Tech Force.
The Mythos context: In May 2026, Anthropic launched its Mythos model, a.k.a. Project Glasswing. Their announcement blogpost read: “Claude Mythos Preview is a general-purpose, unreleased frontier model that reveals a stark fact: AI models have reached a level of coding capability where they can surpass all but the most skilled humans at finding and exploiting software vulnerabilities.”
At the time of writing this report, over 150 organisations across 15 countries have access to Mythos. India is not part of those countries.
How did India react to Anthropic’s Mythos?
- Unprecedented threat: On April 23, Finance Minister Nirmala Sitharaman chaired a meeting with Reserve Bank of India (RBI), National Payments Corporation of India (NPCI), Indian Banks’ Association (IBA), and Indian Computer Emergency Response Team (CERT-In) officials, calling the Mythos threat “unprecedented.”
- MeitY’s attempts for Mythos access: On April 28, Ministry of Electronics and Information Technology (MeitY) Secretary S. Krishnan confirmed that the government is still working out logistics with US authorities to secure access for Indian entities. Read more about it here.
- Servers of AI models used in sensitive sectors should be in India: On May 10, government officials from the Finance Ministry, MeitY, and CERT-In reportedly met with Anthropic’s India team to assess cybersecurity risks to critical sectors like banking, telecom, and power infrastructure. Officials appear to have argued that models used in sensitive sectors must be hosted on Indian servers to address regulatory, security, and jurisdiction risks. Read more about it here.
When Mythos was accessed without authorization: On April 22, some unauthorized users accessed Anthropic’s Mythos AI model, raising concerns about potential misuse of a system which could enable dangerous cyberattacks, according to a Bloomberg report. Anthropic confirmed it is investigating the incident in a third-party vendor environment and said it has found no evidence that core systems were affected or the breach extended beyond that vendor.
‘Mythos and the politics of AI’: “A model that can find and exploit zero-days at scale is not only a defensive tool. It is a strategic capability that militaries will want. This explains why the US Department of War went after Anthropic: this is like cyber-nuclear power. Can be used for good (improve software/produce electricity) or destruction (cyberattacks/nuclear weapons). This is where the difference between democracies and authoritarian regimes comes in: if this were in China, then Anthropic would not have been able to refuse giving it to the government,” reads an excerpt from MediaNama’s editor Nikhil Pahwa’s blogpost on Reasoned newsletter. Read the full blog here.
Also Read:
