The White House ordered AI giant Anthropic to restrict the export of its Fable and Mythos models. This applies to anyone outside the U.S. and foreign nationals within the country. The directive cites unspecified national security concerns, according to TechCrunch. A growing government effort to control advanced technology is evident, but effective regulation of rapidly evolving tech remains challenging.
Governments are expanding export controls to cutting-edge technologies like AI. However, history shows these controls are often circumvented or lead to human rights abuses. For instance, the U.S. government investigated PGP's creator in the 1990s for alleged arms export violations. More recently, Germany-based FinFisher shut down in 2022 after allegations of selling spyware without a license, as reported by TechCrunch.
Despite these efforts, sensitive technologies will likely continue to proliferate globally. Constant re-evaluation of enforcement strategies and international cooperation is demanded to mitigate security and human rights risks. A significant policy disconnect is evident.
Export Control Challenges and Failures
Export controls aim to prevent sensitive technologies from falling into the wrong hands, whether for national security or human rights reasons. The White House's order to Anthropic on its Fable and Mythos models exemplifies national security-driven controls on emerging AI capabilities (TechCrunch). Yet, the track record for surveillance technology controls shows consistent failure.
Bulgaria, for example, licensed surveillance equipment exports to repressive regimes between 2018 and 2023, according to Human Rights Watch. The licensing occurred despite EU obligations for member states to consider human rights records when assessing such licenses. The Bulgarian company Circles received licenses for telecommunication interception systems and monitoring software, as documented by Human Rights Watch. The conviction of Circles' founder Tal Dillian in Greece for spyware use further confirms the failure of existing frameworks to prevent human rights abuses. The failure extends beyond rogue actors, encompassing state-sanctioned exports of sensitive technology.
These cases reveal that while intentions vary—from national security to human rights—the practical implementation of export controls consistently faces significant hurdles and unintended consequences. The White House's effort to restrict AI models like Anthropic's Fable and Mythos appears undermined when EU members like Bulgaria license proven surveillance technologies to repressive regimes, as documented by Human Rights Watch. The regulatory disconnect directly contributes to human rights abuses.
The Persistent Regulatory Disconnect
The White House's focus on 'unspecified national security concerns' for AI export restrictions contrasts sharply with documented human rights abuses from poorly regulated surveillance tech exports by U.S. allies. A misaligned global priority in controlling dual-use technologies is indicated. The U.S. restricts future AI capabilities for national security, while European allies fail to control existing surveillance tech exports to regimes with poor human rights records. A dangerous regulatory disconnect is created.
EU member states routinely ignore their legal obligation to consider human rights in export licensing, as Bulgaria's extensive licensing of surveillance tech to problematic nations proves. Bulgaria's extensive licensing directly contributes to human rights violations. The historical challenge of controlling dual-use technologies, from encryption to modern surveillance, persists due to inconsistent international enforcement. New AI controls will likely face similar circumvention.
Given the persistent regulatory disconnect and historical failures in enforcement, new AI export controls will likely struggle to prevent global proliferation and could inadvertently enable further human rights abuses if international cooperation remains inconsistent.
