US Government Halts Anthropic's Fable 5 and Mythos 5 Amid National Security Concerns

In a significant and immediate move that has sent ripples through the artificial intelligence industry, the US government recently ordered Anthropic, a leading AI safety and research company, to abruptly disable access to its two newest and most powerful AI models: Fable 5 and Mythos 5. The directive, issued by the Trump administration's Commerce Department on June 12-13, 2026, cited national security concerns, specifically pointing to an alleged "jailbreak" that could bypass the models' safety guardrails.

The decision came just days after Anthropic rolled out Fable 5, described as a "Mythos-class" model, which many experts considered a new tier of capability, particularly in cybersecurity applications. The government's export control directive mandated that Anthropic suspend access for all foreign nationals, a requirement so broad that the company felt compelled to disable both models for all its customers globally to ensure compliance. This unprecedented intervention highlights the escalating tensions between rapid AI advancement and the urgent need for robust regulatory frameworks.

The Models at the Center: Fable 5 and Mythos 5

Anthropic, founded in 2021 by former OpenAI members, has distinguished itself with a strong focus on AI safety, developing systems that are reliable, interpretable, and steerable. Its flagship products are the Claude series of large language models. Fable 5 was introduced around June 9, 2026, as the first publicly available model in Anthropic's new "Mythos-class" tier, signifying a leap in capability beyond its existing Opus line. Mythos 5, the underlying and even more powerful model, was reportedly restricted to select partners, including government agencies via "Project Glasswing."

Anthropic had positioned Fable 5 as a highly capable cybersecurity model, equipped with guardrails designed to prevent its misuse in risky areas. The company acknowledged the inherent risks of deploying such powerful AI, stating in its release that "[r]eleasing a model this capable comes with risks," and that "Without safeguards, Fable 5's capabilities in areas like cybersecurity could be misused to cause serious damage." These safeguards were intended to block its use for certain sensitive topics.

The Alleged Jailbreak and Amazon's Role

The catalyst for the government's swift action was an alleged "jailbreak" of Fable 5's guardrails. Reports indicate that researchers at Amazon, a major investor in Anthropic, discovered a method to bypass these restrictions. Amazon CEO Andy Jassy reportedly communicated these findings to senior Trump administration officials, including Treasury Secretary Scott Bessent, expressing concerns that Fable 5 could be coerced into identifying software vulnerabilities.

According to Katie Moussouris, CEO of Luta Security, who reviewed a report shared by Anthropic, Amazon researchers used a series of prompts to circumvent Fable 5's guardrails and access the underlying Mythos model. This allowed them to generate output that could be used to find vulnerabilities and create test scripts for potential proof-of-concept code. White House adviser David Sacks further elaborated on the situation, claiming that the administration asked Anthropic to either fix the reported jailbreak or pull the model, and that Anthropic CEO Dario Amodei refused.

The US government's concern was that foreign adversaries could exploit such bypasses to launch sophisticated cyberattacks, particularly in critical sectors like banking. This fear propelled the Commerce Department to issue an export control directive, prohibiting any foreign national, whether inside or outside the US, including Anthropic's own foreign employees, from accessing Fable 5 and Mythos 5.

Anthropic's Response and Disagreement

In response to the government's order, Anthropic issued a statement confirming its compliance but also expressing strong disagreement with the rationale behind the directive. The company stated that its understanding was that the government had received "verbal evidence of a potential narrow, non-universal jailbreak."

Anthropic argued that the identified vulnerabilities were "relatively simple" and that "other publicly available models are able to discover them as well without requiring a bypass." The company emphasized that if such a standard were applied across the industry, it "would essentially halt all new model deployments for all frontier model providers." This stance highlights a fundamental tension in AI safety: the difficulty of achieving perfect jailbreak resistance and the potential for over-regulation to stifle innovation.

Anthropic also noted its ongoing efforts in AI safety research, including its "Constitutional AI" framework aimed at aligning AI systems with human values, and its "Frontier Red Team" which specifically analyzes implications for cybersecurity. The company had even implemented a 30-day data retention policy for Fable 5 customers to aid in researching and mitigating jailbreaks, incurring "real costs" in the process.

The Cybersecurity Community's Reaction

The US government's drastic measure quickly drew criticism from a significant portion of the cybersecurity community. Dozens of prominent cybersecurity experts signed an open letter to Commerce Secretary Howard Lutnick and National Cyber Director Sean Cairncross, blasting the ban as "dangerous" and "misguided."

These experts argued that the ban had "taken the best models away from defenders, created market uncertainty, and risked America's AI leadership without any real risk to justify it." They pointed out that the ability to uncover cybersecurity vulnerabilities is not unique to Anthropic's models and that American cybersecurity professionals now lack access to powerful tools that could be used for defensive purposes. Some critics, like Katie Moussouris, CEO of Luta Security, directly challenged the government's interpretation of the jailbreak, stating, "That is not a guardrail bypass."

Broader Implications for AI Regulation and Innovation

This incident marks a significant escalation in the US government's efforts to control access to frontier AI models. Historically, export controls have focused on hardware like semiconductor chips, not on the AI models themselves. The directive on Fable 5 and Mythos 5 sets a new precedent, indicating that the government increasingly views advanced AI models as national security assets.

The dispute also underscores the growing tension between AI developers and regulators regarding how to assess and manage risks from "jailbreaks" and other methods used to bypass model safeguards. While the government aims to prevent foreign adversaries from leveraging American AI capabilities, critics worry about the chilling effect this could have on innovation.

Brad Carson, president of Americans for Responsible Innovation, commented that "AI regulation will be some of the most serious and consequential work the U.S. government does over the next generation, and it is imperative that this work be done consistently across industry, without favor, and according to a clear, rules-based process." He added that based on current information, "the decision to block Anthropic's latest AI model fails that test, and as a result, risks America's edge in AI innovation."

The situation also feeds into the ongoing global competition for AI supremacy, particularly between the US and China. The European Union's reaction, with statements on the need for "technological sovereignty" after gaining access to the Mythos platform earlier in June 2026, further highlights the geopolitical dimensions of AI development and regulation.

The Underlying Tensions

This ban is not an isolated incident but rather comes amidst a contentious relationship between Anthropic and the Trump administration. Earlier in the year, Anthropic's relationship with the government reportedly "ruptured" after the company refused to allow the US military to use its AI models for domestic surveillance and fully autonomous weapons systems. This refusal led to the Pentagon placing Anthropic on a supply chain blacklist, an action that was temporarily enjoined by a federal judge.

The current directive, therefore, can be seen in the context of these broader disagreements over the ethical and military applications of advanced AI. The government's swift and broad action suggests a heightened level of concern, whether technically justified or politically motivated, regarding the dual-use nature of cutting-edge AI technologies.

As the AI landscape continues to evolve at an unprecedented pace, the Anthropic Fable 5 and Mythos 5 ban serves as a stark reminder of the complex challenges involved in balancing national security, technological innovation, and responsible AI development. The industry, policymakers, and the public will be closely watching how this situation unfolds and what it means for the future of AI regulation worldwide.

Frequently Asked Questions

What are Anthropic's Fable 5 and Mythos 5 models?

Anthropic's Fable 5 and Mythos 5 are advanced artificial intelligence models. Fable 5 was a publicly released model in Anthropic's new "Mythos-class" tier, known for its powerful cybersecurity capabilities. Mythos 5 is the underlying, even more capable model, primarily accessible to select partners.

Why did the US government ban Fable 5 and Mythos 5?

The US government, specifically the Trump administration's Commerce Department, issued an export control directive banning Fable 5 and Mythos 5 due to national security concerns. The ban stemmed from reports, notably by Amazon researchers, of a "jailbreak" that could bypass Fable 5's safety guardrails, allowing it to identify software vulnerabilities, which the government feared could be misused for cyberattacks.

How did Anthropic respond to the ban?

Anthropic complied with the government's directive, disabling Fable 5 and Mythos 5 for all customers globally due to the broad nature of the export control order. However, the company publicly disagreed with the government's rationale, stating that the alleged jailbreak uncovered only minor, known vulnerabilities and that similar capabilities exist in other publicly available AI models.

What are the broader implications of this ban for the AI industry?

The ban sets a significant precedent by restricting access to AI models themselves, not just the hardware that powers them, signaling an escalation in government oversight of frontier AI. It highlights the ongoing challenges of AI safety, guardrail effectiveness, and the balance between national security and technological innovation. Many experts fear it could stifle AI development and hinder the use of powerful AI tools for defensive cybersecurity purposes.