AI safety debate links coordination, regulation and competition with China

An analysis argues that AI developers face a collective-action problem that may require coordination, while voluntary agreements could need assurances from the US government that they would not violate antitrust laws. It proposes targeted regulation focused on corporate transparency and independent third-party evaluations with access to relevant internal information.

The analysis says a Chinese lead in AI could pose a grave danger to the United States and the world. It supports restricting China’s access to powerful AI chips and semiconductor equipment, as well as strengthening security measures inside AI companies to prevent the theft of key data.

It remains unclear whether the risks of unchecked AI development can be reduced without weakening the US lead over China. Resistance to data centers could potentially expand into a broader coalition for slowing AI, but the distinction between slowing infrastructure construction and slowing AI training and capability development may be too nuanced for political debate.