TLDR
- Anthropic restored access to Fable 5 globally after the US lifted export controls on June 30.
- Anthropic resumed Mythos 5 for approved US organizations following earlier government clearance.
- The US imposed export restrictions on June 12 due to security concerns around advanced AI models.
- Anthropic suspended both models worldwide because it could not verify user nationality in real time.
- Amazon researchers identified a jailbreak that exposed software vulnerabilities through manipulated prompts.
Anthropic restored access to its advanced AI models after U.S. authorities removed export restrictions imposed in June. The decision allows global rollout of Fable 5 and limited U.S. access to Mythos 5. The move follows security concerns and signals closer coordination between Anthropic and government agencies.
Export Controls Reversal Enables Model Return
Anthropic confirmed that regulators lifted export controls affecting its latest models on June 30. As a result, Anthropic resumed Fable 5 access globally starting July 1. Meanwhile, Anthropic also reinstated Mythos 5 access for approved U.S. organizations.
Previously, the government enforced restrictions on June 12 due to national security concerns. However, Anthropic could not verify user nationality instantly, which created compliance risks. Therefore, Anthropic suspended both models worldwide to avoid regulatory violations.
The controls limited access to foreign nationals under export laws tied to sensitive technologies. As conditions changed, Anthropic aligned with regulators to restore controlled availability. This shift highlights how policy decisions directly affect frontier AI deployment.
Security Findings Triggered Initial Restrictions
Anthropic faced scrutiny after researchers identified a jailbreak vulnerability in Fable 5 safeguards. Specifically, Amazon researchers demonstrated methods to bypass protections and extract sensitive outputs. Consequently, Anthropic responded quickly to contain potential misuse risks.
The jailbreak prompted the model to reveal software vulnerabilities under certain manipulated inputs. In one case, it generated exploit-related code, raising serious cybersecurity concerns. As a result, Anthropic halted access to prevent further exposure.
An Anthropic spokesperson said, “We acted immediately to ensure compliance and user safety.”
The company also reviewed internal safeguards and strengthened model defenses. These actions supported regulatory reassessment and eventual removal of restrictions.
Market Reaction and Policy Coordination Expand
The suspension affected market sentiment tied to Anthropic’s valuation through onchain derivatives trading. Notably, a pre-IPO perpetual contract on Hyperliquid fell about 3.7% after the shutdown. This decline reflected uncertainty around Anthropic’s future listing timeline.
At the same time, Anthropic expanded cooperation with U.S. government agencies on AI oversight. The company now provides early access to frontier systems for designated institutions. Additionally, Anthropic shares threat intelligence and aligns on safety benchmarks.
Anthropic also works with Amazon, Microsoft, and Google to define jailbreak risk scoring standards. These efforts connect with a June 2 executive order on AI security policies. Consequently, Anthropic strengthens ties between innovation, regulation, and national security priorities.
