Statement: US Government Directive to Suspend Access to Fable 5 and Mythos 5 (English-Chinese Bilingual Version)

*Statement on the US Government Directive to Suspend Access to Fable 5 and Mythos 5*

Source: Anthropic official announcement (Announcements) | June 12, 2026 | English-Chinese bilingual version

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Anthropic models will not be affected.

The U.S. government, invoking national security-related authority, has issued an export control directive requiring suspension of all access to Fable 5 and Mythos 5 by foreign nationals, regardless of whether they are in the U.S. or abroad, including foreign Anthropic employees. The practical effect of this order is that, for compliance reasons, we must immediately disable Fable 5 and Mythos 5 for all customers. Access to all other Anthropic models is unaffected.

We received the directive from the government today at 5:21pm (ET). The letter did not provide specific details of its national security concern. Our understanding is that the government believes it has become aware of a method of bypassing, or "jailbreaking" Fable 5. We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities. These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well without requiring a bypass.

We received the directive from the government today at 5:21 p.m. ET. The letter did not provide specific details about the government’s national security concern. Our understanding is that the government believes it has identified a method of bypassing, or “jailbreaking,” Fable 5. We reviewed a demonstration showing this specific method being used to identify a small number of previously known minor vulnerabilities. These vulnerabilities appear to be relatively simple, and we found that other publicly available models can discover them as well without any bypass.

1. Anthropic's Posture on Fable's Safeguards

Anthropic's posture with respect to Fable's safeguards, as laid out in our launch blog post, is the following:

As set out in our launch post, Anthropic's posture on safeguards for Fable is as follows:

• We have instituted strong safeguards that greatly reduce the likelihood that Fable is misused for tasks related to cybersecurity (among others). In fact, our safeguards are so strong that many users have complained that they are overly broad.
• In the weeks leading up to the launch of Fable, Anthropic worked with the US government, the UK AISI, multiple private third-party organizations and internal teams to red-team Fable's safeguards for thousands of hours in total.
• These tests showed that Fable's safeguards are substantially more effective than those of any previously deployed model.
• No testers have yet been able to find a universal jailbreak—a jailbreak method that can very broadly bypass the model's safeguards, unblocking a wide range of cyber capabilities.
• We suspect that perfect jailbreak resistance is not currently possible for any model provider. Every safeguard used in the industry is vulnerable to non-universal jailbreaks (which can elicit some cyber information in specific circumstances), and it is likely that universal jailbreaks will eventually be found in the future. We stated this clearly when we released Fable 5.

• We have put in place robust safeguards that significantly reduce the chance that Fable is misused for cybersecurity-related tasks (among other uses). In fact, our safeguards are so strict that many users have complained that they are too broad.
• In the weeks before Fable launched, Anthropic worked with the U.S. government, the UK AISI, several private third-party organizations, and internal teams to conduct red-team testing of Fable’s safeguards for a total of thousands of hours.
• These tests showed that Fable’s safeguards are substantially more effective than those of any model previously deployed.
• No testers have yet found a universal jailbreak—a jailbreak method that broadly bypasses the model’s safeguards and unlocks a wide range of cyber capabilities.
• We believe perfect jailbreak resistance is not currently achievable for any model provider. Every safeguard currently used in the industry is vulnerable to non-universal jailbreaks (which can extract some cybersecurity information in specific circumstances), and it is likely that universal jailbreaks will eventually be discovered in the future. We stated this clearly when we launched Fable 5.

Given that perfect jailbreak resistance does not appear to be possible today, Anthropic adopted a defense in depth strategy with Fable 5. We aimed to make jailbreaks either narrow (in the case of non-universal jailbreaks) or very expensive to produce (in the case of universal jailbreaks), and to combine this with thorough monitoring to quickly detect and shut down any successful attacks. This is also why Anthropic has required 30-day retention of customer data with Fable—a policy change that carries real costs for us with customers, but that allows us to research and mitigate jailbreaks.

Given that perfect jailbreak resistance does not appear possible today, Anthropic adopted a defense-in-depth strategy for Fable 5. The goal was to make jailbreaks either narrow in scope (for non-universal jailbreaks) or very costly to produce (for universal jailbreaks), and to back this with thorough monitoring that can quickly detect and shut down successful attacks. This is also why Anthropic requires 30-day customer data retention with Fable. This policy change does impose real costs on us and our customers, but it enables us to investigate and mitigate jailbreaks.

We stand by this defense in depth strategy. It reduces the risks posed by Fable, making them comparable to the risks of existing models already deployed across the industry.

We stand by this defense-in-depth strategy. It reduces the risks associated with Fable, bringing them in line with the risks of existing models already deployed across the industry.

2. The Facts on Jailbreak Disclosures

We have not even received a disclosure of a concerning non-universal potential jailbreak that led to a harmful result. The potential jailbreaks that have been disclosed to us are either entirely benign responses or are minor findings that provide no Mythos-specific uplift.

We have not received any disclosure of a concerning non-universal potential jailbreak that led to a harmful outcome. The disclosed potential jailbreaks are either wholly benign responses or minor findings that provide no Mythos-specific enhancement.

To date, the government has only given us verbal evidence of a potential narrow, non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws. Our understanding is that one potential jailbreak was shared with the government. We have reviewed a report that we believe is the basis of the government's directive and validated that the level of capability displayed there is widely available from other models (including OpenAI's GPT-5.5), and is used every day by the defenders who keep systems safe. We will share more details over the next 24 hours.

To date, the government has only provided verbal evidence of a potential narrow, non-universal jailbreak, which essentially amounts to asking the model to read a particular codebase and patch security issues in the software. We understand that one potential jailbreak was shared with the government. We have reviewed what we believe is the report underlying the government directive and confirmed that the level of capability shown there is broadly available in other models (including OpenAI’s GPT-5.5) and is used every day by defenders who keep systems secure. We will share more details over the next 24 hours.

3. Compliance, Disagreement, and Apology

We are complying with the government's legal directive and are removing access to Fable 5 and Mythos 5 for all users. However, we disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.

We are complying with the government’s legal directive and removing access to Fable 5 and Mythos 5 for all users. At the same time, we disagree that the discovery of a narrow potential jailbreak should justify the recall of a commercial model deployed to hundreds of millions of people. If this standard were applied across the industry, we believe it would effectively stop all new model launches by frontier model providers.

As we have stated publicly, we believe the government should have the ability to block unsafe deployments, as part of a statutory process that is transparent, fair, clear, and grounded in technical facts. This action does not adhere to those principles.

As we have stated publicly, we believe the government should have authority to block unsafe deployments as part of a statutory process that is transparent, fair, clear, and grounded in technical facts. This action does not conform to those principles.

We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.

We apologize for the disruption this causes to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.

*This article is based on Anthropic’s official announcement. The English-Chinese bilingual translation was prepared by Lamjin for reference only. If there is any discrepancy between this translation and the English original, the English original prevails.*

If you encounter access issues or account limitations when using Claude or Fable 5, we recommend referring to “AI Account Risk Control Checklist: IP Reputation and Network Environment Diagnostics” to troubleshoot network environment issues.

*This article is based on Anthropic’s official announcement, and the bilingual translation was compiled by Lamjin for study and reference only. If there is any discrepancy with the English original, the English original takes precedence.*

Frequently Asked Questions

What is Fable 5?

Fable 5 is a high-capability AI model under Anthropic, comparable in level to Claude Opus, and by naming convention at publication time it is positioned as a next-generation flagship offering. Mythos 5 is another model launched at the same time. The government directive discussed in this article suspends access to both models for all foreign nationals.

How does this ban affect ordinary users?

During the effective period of the directive, all foreign nationals—including foreign employees located in the United States—will not be able to access Fable 5 and Mythos 5. Access to other Anthropic models (such as Claude Sonnet and Claude Haiku) is not affected.

What is jailbreak?

A jailbreak refers to bypassing a model’s safeguards via specific prompts or technical methods so it produces content that it would otherwise refuse. The directive here refers to a "narrow non-universal jailbreak," meaning a workaround that bypasses certain restrictions only under specific conditions and is not a universal jailbreak that removes protections across the board.

What is Anthropic's stance?

Anthropic says it is complying with the legal directive and removing access, but does not accept that this standard should be used to recall a commercial model. It argues that if applied industry-wide, this would effectively halt all new deployments of frontier models. Anthropic also maintains that government oversight should follow a transparent, fair, and technically grounded statutory process.

References

• Anthropic Announcements (official announcement page)