AI safety fears dominate U.S. midterm race as tech leaders clash on slowdown
AI alarm over advanced models is being used by politicians ahead of the midterm elections, with divergent calls for a development pause.
In the week leading up to the U.S. midterms, a spate of AI mishaps reported by OpenAI and a call from Anthropic CEO Dario Amodei for a coordinated slowdown have heightened alarm over frontier AI. Elon Musk echoed the slowdown plea, whereas Meta’s Mark Zuckerberg and Nvidia’s Jensen Huang said it isn’t needed; OpenAI’s Sam Altman also postponed its IPO, seeking a trillion-plus valuation instead. Politicians have seized the moment: Bernie Sanders and Steve Bannon jointly warned about AI, with Sanders advocating an outright ban on superintelligent systems and urging a pause with China, while Bannon pushed for rapid chip restrictions and White House leadership.
President Trump dismissed safety fears as a hoax and emphasized beating China in the AI race. Polls reveal a majority of voters oppose new data centers and are more concerned than excited about AI, and the House has only passed legislation shielding ratepayers from data-center electricity costs, leaving AI safety untouched as the election approaches.
Why it matters
Voter attitudes and legislative action on AI could shape U.S. technology policy and competitiveness before the midterms.
How this story developed
- Jul 28 AI leaders urge U.S. to back global framework for slowing automated AI progress
- Sep 15 President Donald Trump rejected the CEOs' appeals and senior Republicans signaled reluctance, while Senator John Kennedy introduced a “kill switch” proposal.
- Sep 15 China’s security ministry released a notice warning that generative AI tools are being weaponised by hostile forces.
- Sep 15 Trump and Xi are unlikely to agree to an AI slowdown at the upcoming summit.
- Sep 16 While a recent poll shows two-thirds of Americans fear AI could end humanity, conference goers largely dismissed imminent extinction. Attendees such as Dean Levita and Henry Sandles said they trust safeguards and focus on labor displacement, while Sara Jamison and Ashleigh Kaye highlighted the environmental toll of data centers. Others, like Deep Karia and Soumya Nimmu, worried about privacy, mental-health strain and misuse of AI in deepfakes and weapons.
- Sep 16 Chinese Defense Minister Dong Jun called for multilateral AI cooperation at the Xiangshan Forum and US Treasury Secretary Scott Bessent testified on maintaining America’s tech edge.
- Sep 16 Jacob Coxon resigned from Anthropic after issuing the warning.
- Sep 17 In response to recent high-profile incidents involving rogue AI agents, CEOs and founders such as Dario Amodei, Sam Altman, Demis Hassabis, Satya Nadella and Elon Musk signed a joint statement calling for slower progress. The proposal outlines three steps: third-party safety auditors, industry coordination on standards, and global cooperation among democratic governments. Companies like Anthropic have already pledged to allow external evaluators access to their models.
- Sep 17 President Donald Trump met President Xi Jinping in Washington to discuss AI safety.
- Sep 18 Mark Zuckerberg and Jensen Huang publicly opposed a unified slowdown, emphasizing market incentives.
- Sep 18 Lawmakers have begun focusing on updating Section 230 and school‑board decisions related to AI.
- Sep 19 In a lengthy Truth Social post, Trump said the AI Force will mirror the Space Force and aim to protect the fast-growing AI sector. He promised the initiative will encourage development while using existing laws to curb abuses, and he pledged not to hinder the industry’s growth. The post also claimed AI could become as significant as the Internet, potentially representing a quarter of the U.S. economy, and asserted that the United States currently leads China in the field.
- Sep 19 Trump added the AI Force proposal to his agenda.
- Sep 20 Treasury Secretary Scott Bessent met Vice Premier He Lifeng in New York to discuss AI safety and trade ahead of the Trump‑Xi summit.
In this story
Related stories
64 in this thread