Central banks from Tokyo to Frankfurt activated crisis protocols this week after Anthropic unveiled Mythos, an artificial intelligence system whose capabilities have prompted what one G7 official called "the most serious technology assessment since nuclear proliferation."

The emergency response centers not on what Mythos can do — Anthropic has released limited technical details — but on who gets to use it. The San Francisco-based AI company has implemented unprecedented access restrictions, essentially positioning itself as gatekeeper to a technology that financial regulators and security services believe could reshape markets, cryptography, and intelligence operations.

According to the New York Times, which first reported the global response, at least fourteen central banks have established task forces to assess potential economic impacts. The Federal Reserve, European Central Bank, and Bank of Japan held a joint emergency video conference on Monday, the first such trilateral meeting outside of a financial crisis since 2008.

The Access Question

Unlike previous major AI releases from OpenAI, Google, or Meta — which typically become available through APIs or direct downloads — Anthropic has granted Mythos access to fewer than thirty organizations worldwide. The selection criteria remain opaque.

"We're in uncharted territory," said Dr. Sarah Mendez, former chief economist at the IMF and now a senior fellow at the Brookings Institution. "A private company is making sovereign-level decisions about who can access a technology that could potentially affect monetary policy, market stability, and national security. There's no regulatory framework for this."

The Bank for International Settlements, often called the central bank for central banks, issued a rare public statement Tuesday expressing "serious concern" about the "asymmetric distribution of transformative computational capabilities." The carefully worded release stopped short of calling for regulation but noted that "financial stability depends on a level playing field among market participants."

What Makes Mythos Different

While Anthropic has not published a technical paper on Mythos — itself a departure from AI industry norms — briefings provided to select government officials suggest the model represents a fundamental architectural breakthrough rather than incremental scaling.

Three sources familiar with the classified briefings, speaking on condition of anonymity, indicated that Mythos demonstrates "emergent capabilities" in mathematical reasoning, cryptographic analysis, and complex system modeling that exceed previous benchmarks by orders of magnitude.

One capability reportedly demonstrated to U.S. Treasury officials: the ability to identify previously unknown correlations in global financial data that could predict market movements hours or days in advance. If accurate, such capability would effectively render traditional high-frequency trading obsolete and could allow holders to extract enormous profits from markets — or destabilize them entirely.

"The question isn't whether this technology is powerful," said James Chen, director of the AI Security Initiative at Stanford. "The question is whether we're comfortable with Anthropic's board of directors having more influence over global finance than most finance ministers."

Intelligence Community Response

The response from intelligence agencies has been equally swift but far more secretive. According to the Times, the Five Eyes intelligence alliance — comprising the U.S., U.K., Canada, Australia, and New Zealand — held an emergency director-level meeting last week, only the fourth such gathering in the alliance's history outside of active military conflicts.

The concern extends beyond financial markets. Advanced AI systems with breakthrough capabilities in cryptography could theoretically compromise encrypted communications that protect everything from diplomatic cables to military commands to banking transactions.

Israel's signals intelligence unit, Unit 8200, reportedly requested access to Mythos within hours of its announcement. The request was denied. So was a similar request from Britain's GCHQ. Both agencies declined to comment.

"We're seeing a inversion of the traditional relationship between governments and technology companies," noted Rebecca Torres, who served as deputy national security advisor in the previous administration. "Historically, the most sensitive technologies were government-controlled and selectively shared with private contractors. Now we have private companies controlling the most sensitive technologies and selectively sharing with governments."

Anthropic's Rationale

In a blog post published alongside the Mythos announcement, Anthropic CEO Dario Amodei defended the restricted access approach as "responsible scaling." The post argued that releasing such a powerful system openly could create "catastrophic risks" ranging from market manipulation to the acceleration of biological weapons development.

"We believe the right approach is careful, monitored deployment to organizations with robust security practices and alignment with democratic values," Amodei wrote. The post did not specify how Anthropic defines "democratic values" or who makes those determinations.

The company has previously positioned itself as the safety-focused alternative in the AI race, emphasizing "constitutional AI" and interpretability research. But critics argue that Mythos represents a shift from safety advocacy to power consolidation.

"Anthropic has essentially appointed itself the global AI council," said Dr. Yuki Tanaka, director of the Digital Rights Observatory in Berlin. "They're making civilizational-level decisions behind closed doors. That's not safety — that's oligarchy."

Market Implications

Financial markets have responded with characteristic volatility. Anthropic's valuation in private markets has reportedly tripled to over $300 billion in the week since Mythos was announced, making it more valuable than most central banks' foreign currency reserves.

Meanwhile, shares of competing AI companies have tumbled. If Mythos truly represents a breakthrough rather than incremental progress, billions in AI infrastructure investments by Google, Microsoft, and others could be rendered obsolete overnight.

Hedge funds with Mythos access — reportedly including Bridgewater Associates and Renaissance Technologies — have seen unusual trading patterns, according to analysis by market surveillance firm Eventus. The patterns suggest possible use of predictive capabilities, though no rules currently prohibit AI-enhanced trading at this level.

What Happens Next

Regulatory responses are beginning to take shape. The European Parliament has called an emergency session to consider whether Mythos falls under the EU's AI Act, which wasn't designed to address access restrictions by private companies. U.S. senators from both parties have requested classified briefings from the intelligence community.

Senator Maria Alvarez (D-CA), chair of the Senate Intelligence Committee, issued a statement calling for "immediate hearings on whether critical AI systems should be designated as strategic infrastructure subject to government oversight."

But regulation faces a fundamental challenge: most legislators don't fully understand what Mythos can do, and Anthropic isn't telling them. The company has offered classified briefings to select officials but required non-disclosure agreements that prevent public discussion of capabilities.

"We're trying to regulate something we're not allowed to understand," said one congressional staffer involved in AI policy, speaking anonymously. "It's like trying to write nuclear policy when the only people who know how bombs work won't explain it."

For now, the world's central banks are preparing for scenarios they can only guess at. The Federal Reserve has reportedly begun stress-testing how markets might react if Mythos-equipped firms gain systematic trading advantages. The Bank of England is examining whether AI-driven market prediction constitutes market manipulation under existing law.

The answer to that question — and dozens of others raised by Mythos — will help determine whether the age of transformative AI is governed by democratic institutions or by the private companies that build the systems.

As Dr. Mendez put it: "We're watching in real-time as the balance of power shifts from governments to a handful of AI labs. The question is whether we'll do anything about it before that shift becomes permanent."