AI
Industry pushes safety standards as Trump dismisses AI guardrails
AI researchers and leaders are advocating for voluntary safety measures and coordinated regulation, but face skepticism from the Trump administration and the challenge of keeping pace with Chinese competitors.
Ars Technica
Top figures in artificial intelligence are making the case for a deliberate slowdown in development, coupled with safety measures that would apply across the industry. Yet their calls for caution face an uphill battle—both from the political establishment and the reality of global competition.
What the AI leaders are asking for#
Anthropic CEO Dario Amodei has laid out a vision in which major AI firms would adopt uniform safety standards, ideally enforced through regulation targeting any frontier AI company that refuses to comply voluntarily. The aim is to prevent cutting corners as capabilities advance. However, achieving this would require coordination not just within the US but across democratic nations willing to adopt similar rules.
Frontier AI developers are also being urged to embrace self-regulation. David Sacks, co-chair of the president's Council of Advisors on Science & Technology, wrote on social media that "the easiest way not to build superintelligence is for you to agree not to build it."
The political resistance#
The current administration shows little appetite for mandating AI safeguards. President Trump stated on social media that the only guardrail AI needs is "a STRONG AND SMART (High IQ!) PRESIDENT," suggesting regulatory frameworks are unnecessary.
House Speaker Mike Johnson has struck a more moderate tone, acknowledging that safety measures are needed to prevent AI from spiraling out of control. Yet he has also cautioned against panic or emergency measures, preferring that the industry take the lead rather than Congress imposing restrictions.
The China challenge#
Even if the US and its allies could align on safety standards, there remains a fundamental problem: open-weight models developed in China are advancing rapidly, trailing leading US systems by only a few months. Amodei has proposed escalating levels of international cooperation, ultimately including a global pause on AI development—but he acknowledges such an agreement is unlikely to happen soon. The stakes are too high; any country that breaks ranks risks losing geopolitical advantage. A Chinese defection from such an accord could theoretically grant them dominance in a transformative technology, making collective action extraordinarily difficult to enforce.
