Citing national security concerns, the US government has ordered Anthropic to halt access to its newest AI models, Fable 5 and Mythos 5. The decision is seen as a major step in regulating advanced AI systems and could have global repercussions, including for users in India.
Anthropic confirmed on Friday that it has received a US government order to restrict access to Fable 5 and Mythos 5, preventing foreign nationals in the US and abroad from using the models.
In its statement, the company said the order forces it to immediately cut access to Fable 5 and Mythos 5 for all customers, adding that all other models will continue to be available.
The instruction reportedly includes both customers and foreign national staff working at the company. The suspension was announced only days after Anthropic introduced Fable 5, a commercial version derived from the more powerful Mythos 5 system that had been heavily restricted for cybersecurity reasons.
Anthropic said the government has reportedly discovered a technique that may circumvent certain safety protections in Fable 5. The firm said it was informed via a letter from the government on June 12. Although the letter did not specify the nature of the national security concern, Anthropic said officials appear to fear a “jailbreak” method that could enable exploitation of the model’s capabilities.
It comes in the wake of an executive order signed by US President Donald Trump earlier this month, creating a system to review advanced AI models for potential national security risks prior to release. The directive underscores increasing worry in Washington that powerful AI models could be used for cyberattacks, intelligence operations, or other security threats.
Mythos is seen as one of Anthropic’s most powerful AI systems, having demonstrated advanced abilities in coding tasks, cybersecurity, and autonomous reasoning. Anthropic has previously limited access to Mythos because of concerns that its capabilities could be misused by malicious actors.
Fable 5 was designed as a more accessible alternative, but it still preserves many of the advanced features that attract enterprise and developer users to Mythos. The concern is that these capabilities might be repurposed for cyberattacks if existing safeguards are successfully bypassed.
Anthropic noted that authorities have not offered proof of a significant or systemic security failure. The company said the so-called jailbreak was narrow and restricted, and could only surface a few already known software vulnerabilities. Anthropic added that similar findings can be made using other public AI models without circumventing safeguards.
According to the company, Fable 5 was tested for weeks with US government agencies, the UK AI Safety Institute, and independent researchers prior to release, with no universal jailbreak discovered that could widely bypass its protections. According to Anthropic, the situation appears to be a “misunderstanding,” and the company is trying to restore access at the earliest.
The decision could impact India significantly, a key and fast-growing market for Anthropic. The firm has recently teamed up with Tata Consultancy Services (TCS) to roll out access to its AI models for tens of thousands of staff. Infosys, too, has announced partnerships built around Anthropic’s AI for enterprise solutions.
For breaking news and live news updates, like us on Facebook or follow us on Twitter and Instagram. Read more on Latest Technology News on India.com.
Hritika Mitra is a Senior Sub-Editor at India.com. In her four years-long career, she has covered events ranging from the Iran-Israel War, the 2024 US presidential election, and the Russia-Ukraine War … Read More
Subscribe to Our Newsletter Today!
By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts Cookies Policy.
© 1998-2026 INDIADOTCOM DIGITAL PRIVATE LIMITED, ALL RIGHTS RESERVED
Leave a Reply