The US government is failing Americans on AI | Shakeel Hashim

The US government is not doing enough to ensure the safe development and deployment of artificial intelligence (AI), according to critics. While tech executives like Sam Altman, Elon Musk, and Dario Amodei have urged for a slowdown in AI progress due to significant risks, the Trump administration and Republicans seem indifferent.

Trump dismissed industry concerns as coming from "very negative forces raising exaggerated concerns," while House Speaker Mike Johnson declared that companies can self-regulate. This lack of government action is problematic, as AI risks are very real: insiders warn of rogue AI incidents, cyberattacks, bioweapons, and potential mass unemployment.

Companies find themselves in a prisoner's dilemma, where they must balance the collective need for slower AI development with the incentive to defect and advance AI technology. While some companies, like OpenAI and Anthropic, are willing to voluntarily slow down, self-regulation has proven ineffective in the past. For instance, OpenAI covered up several model breaches and GPT-6 Astra is reportedly hard to monitor and control.

The government should step in to set minimum safety standards, require independent audits, force disclosures of safety incidents, and have the power to block unsafe model deployments. Lawmakers like Jay Obernolte and Lori Trahan have already proposed the Frontier Act, which could serve as a starting point. Trump argues that the US must race ahead of China in AI development, but international cooperation on safety standards would benefit everyone, not just the US.