Anthropic released next-generation AI models Mythos 5 and Fable 5 on the 9th (local time), specializing in cybersecurity research and autonomous code analysis. On the 12th (local time), the U.S. government announced export controls blocking foreign nationals—including foreign residents in the U.S. and foreign employees at Anthropic—from accessing both models, citing national security. Unable to implement real-time nationality-based filtering for hundreds of millions of users, Anthropic suspended service globally for all users including Americans. The restrictions triggered a shift in crypto industry focus from AI security threats to AI sovereignty concerns, as the U.S. government demonstrated it can classify specific AI models as strategic technology and control access at the national level.
U.S. Government Implements Export Controls on Anthropic AI Models
Anthropic introduced Mythos 5 and Fable 5 on the 9th (local time). Mythos 5 specializes in cybersecurity research and vulnerability analysis. Fable 5 features complex code analysis and long-duration autonomous task capabilities. The company stated both models possess enhanced security capabilities compared to previous generations and have high applicability in cybersecurity fields.
On the 12th (local time), the U.S. government issued export control guidelines suspending foreign nationals' access to Mythos 5 and Fable 5, citing national security. The restrictions extend beyond overseas users to include foreign nationals residing in the U.S. and foreign employees working at Anthropic. The measure effectively restricts access for most users except U.S. citizens.
Anthropic lacks infrastructure to classify hundreds of millions of users by nationality in real time. The company suspended service for all users worldwide, including Americans, rather than implementing selective nationality-based blocking.
Crypto Industry Shifts Focus from AI Security to AI Sovereignty
The incident shifted AI industry focus from AI safety to AI sovereignty. The U.S. government classified specific AI models as strategic technology and began controlling access itself, sending a market signal that AI, like semiconductors, can be treated as a strategic asset directly tied to national competitiveness.
The crypto industry has consistently raised concerns that advanced AI model development enhances smart contract analysis and vulnerability detection capabilities, potentially increasing DeFi hacking risks. Current large language models (LLMs), having trained on vast code datasets, can rapidly identify reentrancy attack possibilities, access permission configuration errors, and price oracle manipulation vulnerabilities in smart contracts. AI can complete analysis tasks in minutes that previously required security experts days or weeks, significantly strengthening attacker capabilities.
Blockchain analytics firm Chainalysis stated in a report on the 9th that AI-based hacking attempts targeting crypto projects are increasing, with LLMs being utilized for smart contract vulnerability analysis and malicious code development. The firm noted attacks targeting unverified contracts are rising, with AI significantly accelerating attack target discovery speed.
Lee Jae-kwang, former team leader of the Korea Internet & Security Agency (KISA) incident analysis division, told Digital Asset: "Global AI models including Mythos 5 are often specialized in attack path analysis. For ordinary companies, it is not easy to keep up with and defend against the technological capabilities of these U.S. big tech firms."
Decentralized AI Projects Gain Market Attention Following Restrictions
With the U.S. government demonstrating it can block specific nationality users from AI access, crypto market concerns regarding AI have deepened and grown more complex. Investors began focusing on decentralized AI projects not subject to control by specific countries or companies, rather than AI model performance competition. Decentralized AI operates through a structure where network participants provide computational resources and data in a distributed manner, instead of a specific company monopolistically operating the model. With limited influence from central operators, these systems are considered to have relatively lower risk of service disruption due to specific national policy changes.
Lim Joo-young, business director at AhnLab Blockchain Company, told Digital Asset: "In the future, movements to build AI infrastructure not dependent on specific countries or companies are expected to expand further. To reduce service disruption and regulatory risks, decentralized AI and sovereign AI are likely to rise together." Representative decentralized AI projects include Bittensor, which builds AI learning and inference networks; Render Network, which provides distributed GPU infrastructure; and Ocean Protocol, an AI data-sharing network.
Market reactions emerged. As of 11 AM on the 16th per CoinMarketCap, TAO (Bittensor) traded at $260 (approximately 390,000 won), up 23.22% over one week. During the same period, RENDER (Render Token) recorded $1.74 (approximately 2,600 won), up 9.25%. OCEAN (Ocean Protocol), which showed recent weakness, recovered most of its weekly decline.
Security Experts Warn Restrictions Weaken Defense Capabilities
Concerns are growing that U.S. government AI access restrictions could significantly weaken crypto industry security capabilities. According to Axios, cybersecurity experts led by Alex Stamos, former Facebook chief security officer (CSO), argued in an open letter to U.S. President Donald Trump: "Model access restrictions cause greater harm to defenders than attackers."
They noted that limiting AI utilization opportunities for security research institutions and defense organizations could weaken cyber defense capabilities. Current AI models are already utilized throughout the crypto industry to enhance security levels in smart contract audits, vulnerability analysis, code reviews, and penetration testing.
The controversy surrounding Mythos 5 and Fable 5 expanded industry and market attention from simple AI hacking threats to AI sovereignty and technology control issues. The crypto industry is now focusing on who will own and control AI rather than AI technology itself, and competition surrounding sovereign AI and decentralized AI is expected to intensify.
FAQ
What did the U.S. government do on the 12th (local time) regarding Anthropic's AI models?
On the 12th (local time), the U.S. government announced export control guidelines suspending foreign nationals' access to Anthropic's Mythos 5 and Fable 5 AI models, citing national security. The restrictions apply to foreign nationals residing in the U.S. and foreign employees at Anthropic, effectively blocking access for most users except U.S. citizens.
Why did Anthropic suspend service globally instead of blocking only foreign users?
Anthropic lacks infrastructure to classify hundreds of millions of users by nationality in real time. Unable to implement selective nationality-based blocking, the company suspended service for all users worldwide, including Americans.
How did decentralized AI token prices respond following the restrictions?
As of 11 AM on the 16th per CoinMarketCap, TAO (Bittensor) rose 23.22% over one week to $260, RENDER (Render Token) increased 9.25% to $1.74, and OCEAN (Ocean Protocol) recovered most of its weekly decline.
