Anamika Dey, editor
By TechSun News Desk | techsunnews.com | June 18, 2026 | Tech / AI / News | 6 min read
| 🚨 DEVELOPING — June 18, 2026: Anthropic’s Fable 5 and Mythos 5 AI models remain offline after the Trump administration forced a global shutdown on June 13. Anthropic and the White House say they are ‘working quickly to resolve’ the situation. Both models have been unavailable to all users including Anthropic’s own employees for five days. |
Here is a sentence you would not have expected to read a year ago: an AI company asked the US government to regulate it more strictly. And then five days later the government shut down that same company’s two most powerful AI models with a 90-minute warning.
That is not a summary of a film. That is what happened to Anthropic the company behind the Claude AI assistant between June 10 and June 13, 2026. And the story is still unfolding right now.
To understand why this matters for AI users, for tech investors, and for anyone who cares about who actually controls the most powerful technology ever built you need the full picture. Here it is.
First — What Is Anthropic and Why Does It Matter?
Anthropic is one of the three most important AI companies in the world right now alongside OpenAI and Google DeepMind. It was founded in 2021 by Dario Amodei and others who left OpenAI, specifically because they were worried about AI safety.
The company makes the Claude family of AI models — which we have covered in detail in our ChatGPT vs Claude vs Gemini comparison. Claude has become one of the most widely used AI tools in business, coding, research and writing — a genuine competitor to ChatGPT.
Anthropic has always positioned itself as the safety-first AI company. Where OpenAI moved fast and figured out safety later, Anthropic built safety into its core identity. That positioning is now at the centre of the biggest AI-government clash since the industry began.
The Timeline How This Escalated So Fast
This did not happen overnight. Here is the full sequence:
| Date | What Happened |
| Early 2026 | Pentagon places Anthropic on defense contractor blacklist companies banned from using Claude |
| March 27, 2026 | Federal judge issues temporary injunction blocking blacklist Anthropic sues Trump administration |
| June 2, 2026 | Trump executive order requires AI companies to submit models for government evaluation voluntary |
| June 10, 2026 | Anthropic CEO Dario Amodei publishes essay calling for serious binding AI regulation |
| June 11, 2026 | Anthropic publishes full Advanced AI Framework regulation roadmap for Washington |
| June 13, 2026 — 1:15pm | White House calls Anthropic gives 90-minute deadline to restrict Fable 5 and Mythos 5 |
| June 13, 2026 — 5:21pm | Commerce Secretary sends letter threatening criminal and civil penalties — Anthropic pulls both models worldwide |
| June 16, 2026 | Anthropic sends top security researcher Nicholas Carlini to Washington for emergency talks |
| June 17–18, 2026 | Anthropic and White House ‘working quickly to resolve’ models still offline |
| ⚠️ The 90-minute detail matters: Giving a company 90 minutes to overhaul its entire global access infrastructure is not a negotiation. The technical complexity of restricting a deployed AI model to US citizens only verifying nationality for millions of users across 190 countries — is genuinely not achievable in 90 minutes. Anthropic chose total shutdown over rushed half-compliance. That choice tells you something. |
What Anthropic Actually Asked For — And What Washington Did Instead
The dispute becomes more notable when viewed alongside Anthropic’s earlier calls for AI regulation.
On June 10 three days before the shutdown Anthropic CEO Dario Amodei published an essay and a full policy framework calling for serious binding regulation of AI. His proposal was thoughtful and specific: require frontier AI companies to test their models publicly, submit to independent evaluation, publish safety reports, and give the government authority to block dangerous deployments.
The key word in his proposal was “carefully designed” regulation with guardrails, based on science, with transparent rules that every company follows equally.
What he got three days later was the opposite: a 90-minute ultimatum, no specific justification beyond ‘national security’, no transparent rules, no equal treatment just a demand applied to Anthropic specifically, while OpenAI and Google continued operating normally.
| 💬 Alex Stamos, former Facebook security chief: “Those rules need to be written down and transparent. That has not happened. There’s nothing that Anthropic or anybody else can look at to say what can and can’t I do. For every American company, the fear is that at any moment you can now, if you run afoul of the administration, be shut down for a completely arbitrary decision.” |
What Was the Actual Security Concern?
The administration cited “national security authorities” but did not specify the concern publicly. According to reports from CNBC and Semafor, the trigger was:
A method to ‘jailbreak’ Fable 5 essentially trick it into ignoring its safety restrictions had been identified. The administration claimed a China-linked group may have accessed the model’s capabilities. Washington’s specific concern was that these advanced models could be used to find and exploit vulnerabilities in critical infrastructure, operating systems and browsers.
Analysts note the situation highlights a broader challenge facing advanced AI systems. Anthropic built Mythos 5 specifically for cybersecurity defenders to help find vulnerabilities before attackers do. Critics argue the same capability that makes it useful for defence makes it dangerous in the wrong hands. The dual-use problem is one that security researchers say the industry has not yet found a clear answer to.
The development reflects broader debates about government oversight of advanced AI systems — including questions about energy infrastructure and military applications that observers say have grown more urgent in 2026.governments around the world trying to control AI they helped fund and enablehow much AI is now consuming in terms of energy and military infrastructure
What Does This Mean for People Who Use Claude?
Right now if you use Claude in any form outside the US you cannot access Fable 5 or Mythos 5. You are being redirected to older Claude models.
Inside the US, US citizens can still access the models. But the disruption has been significant businesses that integrated Fable 5 into their workflows suddenly found it gone, with no warning and no timeline for return.
The disruption illustrates how dependent many businesses have become on third-party AI platforms. Some observers argue that AI tools which appear permanent can be altered or removed overnight whether by commercial decisions or, as in this case, government intervention.AI tools that feel permanent can be removed or changed overnightthe companies building AI tools answer to forces beyond just their users
And if you care about your digital privacy and security which connects to everything from which VPN to use to whether your phone is collecting data on you the question of who controls AI models and what they can be forced to do is directly relevant to you.
The Bigger Picture A New Era of AI Control
Viewed more broadly, the incident raises important questions about AI governance.
An AI company that asked to be regulated that actively wanted government oversight got shut down by executive action with no transparent process, no clear rules and no warning. Its CEO published a thoughtful 5,000-word regulation framework on Tuesday. By Friday his company’s most powerful products were offline worldwide.
Many AI companies are likely to interpret the episode as a signal that government intervention can be swift and far-reaching. Some legal experts argue it does not matter whether a company cooperates with regulators products can be taken offline with little notice if officials cite national security grounds.
Critics argue those earlier debates concerned influence, while this episode represents direct government control of deployed AI systems. Some analysts contend this marks a meaningful shift in the relationship between frontier AI companies and the US government.politicians trying to slow AI data center expansiongovernments wanting equity stakes in AI companies
FAQ
1. Can I still use Claude right now?
Yes, Claude Opus 4.8 and other older models remain available worldwide. Only Fable 5 and Mythos 5 Anthropic’s two most advanced new models released in June 2026 were taken offline. If you were using Claude for writing, coding or research before June 13, those capabilities are still available. The more advanced capabilities introduced in Fable 5 particularly in cybersecurity and complex reasoning are what is currently unavailable outside the US.
2. Is this legal can the government force a company to shut down its AI?
This is genuinely contested legal territory. The administration cited export control authority under national security law authority that is real and broad. Anthropic has filed a lawsuit challenging the earlier Pentagon blacklist, and that litigation is ongoing. Whether the 90-minute ultimatum was lawfully issued under proper procedures is a question that will likely end up before federal courts. The short answer: the government believes it has the authority, Anthropic disputes the process, and courts will eventually decide.
3. What does this mean for OpenAI and other AI companies?
Every AI company is watching this extremely carefully. If the government can shut down Anthropic’s models over unspecified national security concerns, the same authority could theoretically be applied to OpenAI’s GPT models, Google’s Gemini, or any other frontier AI system. The difference so far is that Anthropic has been more willing to publicly disagree with administration policies. Whether companies that stay quiet receive different treatment is a question the industry is quietly but urgently asking right now.
| 💬 What Do You Think? Is the US government right to have this level of control over AI models or does a 90-minute shutdown with no transparent process concern you? And does it change how you feel about using AI tools knowing governments can force them offline at any time? Drop your honest take in the comments this is one of the most important debates in tech right now and we want to hear from real people, not just policy experts. |
techsunnews.com | Tech / AI / News | © 2026
