From PGP to Myth: A Brief History of Export Controls That Stopped No One

Last Friday, the White House ordered Anthropic to restrict exports of its powerful AI models Fable and Mythos to anyone outside the United States and to foreign nationals within the United States, citing unspecified national security concerns. Soon after, the AI ​​giant quickly pulled the plug on both models, which will now be unavailable to anyone for a week.

This episode is the first real test of whether the U.S. government can use export controls to curb frontier AI in the way it has previously tried to curb encryption and spyware. The results are very uneven. And as dramatic as it may sound, how this impasse is resolved could shape not only Anthropic’s access to foreign markets, but the rulebook that other AI labs must build upon.

First, some context. Since Anthropic launched Mythos in April, the company has been marketing it as some kind of Doomsday cyber machine that could wreak havoc on the Internet if released too widely. That’s why before the ban, only about 150 vetted companies and government agencies had access to the system. The goal was to help defenders secure software and services before attackers reach capabilities like Mythos.

So why was the ban imposed? There are reportedly two follow-up events. First: Anthropic gave South Korean telecoms access to Mythos through a limited partner program, and U.S. officials became alarmed after identifying the company as a company suspected of having ties to China. (The company, widely known as SK Telecom, has denied any ties to China.) Amazon CEO Andy Jassy also reportedly alerted the administration after Amazon’s own researchers found a way to circumvent Fable 5’s safeguards. Anthropic disputes the “jailbreak” label, saying it’s not a total defeat of the model’s safety measures, but rather a narrow, already-patched issue.

The result was the same. The Commerce Department issued export control directives, and Anthropic had to scramble to immediately restrict access to its products, in some accounts within about 90 minutes of receiving notification.

But none of this is new. Governments have been using export controls to limit the spread of dangerous cyber technologies for decades, but their track record has been middling at best.

The U.S. government was behind the most dramatic failure of this approach in history in the early to mid-1990s. At the time, computer scientists were developing encryption techniques to protect data traveling over the Internet. One such encryption product is called Pretty Good Privacy (PGP), a popular software that encrypts data and makes it virtually undecipherable even if intercepted while it is being transmitted over the Internet to the intended recipient.

The U.S. government initially viewed PGP as a dangerous weapon. This was due to concerns that PGP might prevent intelligence agencies from snooping on emails as they pass through the wire. In an effort to stop the distribution of PGP, the U.S. Customs Service opened a criminal investigation against PGP’s founder, Phil Zimmermann, for violating arms export controls. He fought back by publishing PGP’s source code in a printed book, sparking what is now known as the “crypto wars.”

Zimmermann later won an important battle when the investigation closed, laying the foundation for critical end-to-end encryption algorithms like those used by billions of Signal and WhatsApp users.

Beginning in the early 2010s, researchers began discovering Western-made spyware being used to target dissidents in the Middle East. In response, several governments agreed to expand the Wassenaar Agreement, an international treaty that restricts exports of dual-use software and technology used for both civilian and military applications.

The idea was to classify surveillance and hacking software as dual-use, so that spyware creators would need an export license to sell their products overseas.

But Wassenaar has always had two inherent weaknesses. There are several countries that do not comply with this agreement, including Israel, which is home to the world’s most active spyware manufacturers.

The agreement also depends on countries applying it to businesses within their borders at their own discretion. The Italian government had for some time granted Hacking Team, one of Italy’s top spyware manufacturers at the time, a license to export its tools globally, despite its history of selling spyware to repressive governments that used it to hack journalists and human rights activists.

Since then, other countries in Europe have loosened up on spyware manufacturers like Italy. Despite numerous scandals, Europe, home to many spyware and hacking tool manufacturers, has consistently failed to curb spyware exports to authoritarian regimes. Critics say recent new efforts across all 27 member states to tackle the growing problem of spyware exports to authoritarian countries “are not going far enough.”

Some spyware manufacturers, such as Intellexa, a consortium of authorized spyware companies, have moved operations to countries with lax export controls. Other spyware manufacturers have also sought to relocate operations to Saudi Arabia for similar reasons.

There were some victories. Germany-based spyware manufacturer FinFisher was shut down in 2022 after a multi-year investigation by German prosecutors into the company for allegedly selling spyware to Turkey without an export license. Investigators previously found FinFisher spyware distributed to the phones of critics of the Turkish government.

At the time of writing, a stalemate remains between Anthropic and the Trump administration. There is a reasonable opportunity for the administration to ease and lift restrictions to keep U.S. AI companies globally competitive. This amounts to a tacit admission that AI labs elsewhere, including in China, are likely to reach similar capabilities regardless of U.S. limitations. Alternatively, U.S. AI companies may need to obtain government approval before serving foreign customers, which could invariably reduce profitability due to compliance burdens.

Given past experience with global governments attempting to control a range of software, government-mandated export controls are unlikely to be the right approach to prevent malicious actors from abusing powerful dual-use cyber technologies.