DW reports on the potential setback to Anthropic's IPO plans and the unusual scope of the export ban, covering both the technical dispute and the deteriorating relationship with the Trump administration.
Technology6 sources analysed
Anthropic blocks AI models on US government order
The US government issued an urgent export control directive to AI company Anthropic, ordering it to block foreign nationals from accessing its newly released advanced AI models, Claude Fable 5 and Claude Mythos 5. Anthropic complied by disabling access for all users globally, citing national security concerns. The order, received at short notice on June 12, is believed to stem from a reported method to bypass Fable 5's safeguards, though Anthropic disputes the severity, arguing the jailbreak is narrow and comparable to capabilities in other models. The incident marks an escalation in the ongoing tensions between Anthropic and the Trump administration, following earlier disputes over military use of AI and a Pentagon supply chain blacklist.
Key Facts
- Anthropic forced to block access to Claude Fable 5 and Mythos 5 for all users after US government export order.
- Order bars foreign nationals—including employees—from using the models, citing national security.
- Government believes a jailbreak method exists for Fable 5; Anthropic says it's narrow and other models have similar capabilities.
- The incident follows a breakdown in relations between Anthropic and the Trump administration over military AI use.
- Anthropic argues the order lacks transparency and could set a dangerous precedent for the entire AI industry.
Source Coverage
Il Fatto highlights CEO Dario Amodei's Italian-American background, the legal battle, and the industry-wide implications if the government's standard is applied broadly. It also notes the lack of transparency.
Al Jazeera reports the order with emphasis on the ban affecting foreigners in the US and the company's disagreement, highlighting the lack of specific details from the government and the potential for the AI to become a cyberweapon.
Engadget focuses on the technical aspects: the models' capabilities (beating Pokémon), the safeguards in place, and Anthropic's defense strategy. It quotes the company’s stated disagreement with the recall standard.
Wired frames the block as a new flashpoint in an ongoing battle between Anthropic and the Trump administration, detailing the prior Pentagon blacklisting and lawsuits, and including a call for tips from insiders.
NOS Dutch news emphasizes the 'super hacker' capability of Mythos, the blanket ban on all non-US entities (including allies), and frames this as a new chapter in the conflict between Anthropic and the US government.
Conclusion
This unprecedented government intervention highlights the growing friction between national security priorities and the commercial deployment of frontier AI. While Anthropic complies under protest, it warns that applying such standards across the industry would halt new model releases. The story underscores the challenges of regulating dual-use AI and the fragile relationship between innovative tech firms and a government wary of foreign access to powerful tools.
Logical analysis
What sources agree on
- Anthropic received a US government directive to block foreign access to Fable 5 and Mythos 5 on June 12.
- Anthropic complied by disabling access for all customers globally.
- The government cited national security concerns, specifically a potential jailbreak method.
- This is part of a broader conflict with the Trump administration over military AI use.
Prior relationship with US government
| Outlet | Claim |
|---|---|
| Wired | Pentagon labeled Anthropic a 'supply chain risk' and Anthropic sued; this is a new flashpoint. |
| NOS | The conflict started because Anthropic refused to allow military use for autonomous weapons; Defense Secretary Hegseth wanted no limits. |
| Il Fatto Quotidiano | The company sued the Pentagon after the blacklisting; the order is a continuation of that legal fight. |
Extent and severity of the jailbreak
| Outlet | Claim |
|---|---|
| Wired | The jailbreak is narrow and non-universal; other models can find the same vulnerabilities. |
| Engadget | The company says the jailbreak is 'potential narrow, non-universal' and that the government's evidence was verbal. |
| Al Jazeera English | The AI is a 'super hacker' that could become a dangerous cyberweapon; the order was based on a report of limited capability but still concerning. |
- Few outlets mention that Fable 5 was specifically designed with safeguards blocking cybersecurity, biology, and chemistry responses; Wired and Engadget do but briefly.
- The specific identity of the entity that reported the jailbreak to the government is not named in any article.
The coverage reveals a consistent narrative of a hurried government order based on a disputed technical assessment. Anthropic's compliance appears cautious but critical of the process. The event illustrates the tension between export control and open AI development. While the government's national security rationale is accepted, the lack of specificity and the broad ban on all foreign nationals—including allies—raises questions about proportionality and could disrupt international AI collaboration. The story is likely to intensify debates over AI governance and the balance between security and innovation.
Related Topics
References
- [1]
- [2]US orders Anthropic to disable AI models for all foreign nationals
Al Jazeera English
- [3]
- [4]
- [5]
- [6]
Get tomorrow's top stories in your inbox
Trending now
Trending