Trump calls AI safety warnings a 'hoax' as industry executives push for mandatory kill switch

Today’s Focus

President Donald Trump on Monday dismissed warnings from AI industry executives about the technology’s risks as a “hoax,” using a series of Truth Social posts to reject calls for new federal safeguards, the BBC reported.

Trump compared the safety concerns to what he called the “Global Warming Scam” and the “RUSSIA, RUSSIA, RUSSIA HOAX,” and declared that the only “guardrails” the industry needed were a “strong and smart” president. He also warned of a “SICK conspiracy going on against AI and Data Centers” and said China was the sole beneficiary of any U.S. slowdown.

The posts followed an interview in which Jack Clark, co-founder of Anthropic, told the BBC that a mandatory third-party-verifiable “kill switch” may be necessary across the industry. Clark’s remarks came alongside a wave of public warnings from staff and executives at leading AI firms about existential and near-term risks from advanced models.

Investors reacted quickly. Shares in several large AI-exposed companies sold off Monday as markets priced in the possibility of tighter oversight, according to the BBC.

Chinese state media seized on the exchange, arguing that “Washington’s growing anxiety over China’s rapid AI development” was distorting U.S. policy debates. Beijing has poured state funding into domestic chip and model development and has framed the competition as a strategic race the U.S. is trying to slow through export controls.

Trump has moved to loosen federal AI oversight since returning to office, rolling back a Biden-era executive order that required safety testing disclosures for the largest models. Congress has not passed comprehensive AI legislation, and a proposed federal preemption of state AI laws stalled earlier this year. The White House has not indicated any new policy will follow Monday’s posts.

The Debate

Supporters argue

Trump and allies in the AI industry frame heavy-handed safety rules as a gift to Beijing. In his posts, the president wrote that “WHOEVER WINS AI, WINS,” casting the contest with China as the overriding concern.

Venture investor Marc Andreessen and other Silicon Valley donors who backed Trump have argued for months that “doomer” narratives are being used by incumbents to lock in market position through regulation. The White House’s AI czar, David Sacks, has repeatedly said on his podcast that mandatory safety testing would slow U.S. model development while Chinese labs like DeepSeek face no equivalent constraints.

Industry groups including NetChoice and the Chamber of Progress said Monday that a patchwork of state-level AI bills already threatens innovation, and that a federal “light touch” is preferable to Europe’s AI Act, which they argue has driven investment away from the EU.

Chinese state media, quoted by the BBC, said U.S. officials “know that China has become a strong competitor in AI,” an argument Trump’s supporters cite as evidence that the safety debate itself has become a strategic vulnerability.

Critics argue

Anthropic’s Jack Clark told the BBC that voluntary industry pledges are insufficient and that a mandatory “kill switch,” verifiable by an independent third party, may be needed as models grow more capable.

Senate Democrats, including Sen. Ed Markey (D-MA), said in a Monday statement that dismissing safety concerns as a hoax was “reckless” and pointed to bipartisan bills requiring pre-deployment testing for frontier models. Sen. Josh Hawley (R-MO), who has partnered with Democrats on AI liability legislation, has argued that “trusting the companies to police themselves has never worked.”

The Center for AI Safety and the Future of Life Institute, which last year organized an open letter signed by hundreds of researchers, said Trump’s comparison to climate denial ignored warnings coming from the industry itself, not outside activists.

Civil-rights groups including the Leadership Conference on Civil and Human Rights said unregulated deployment of AI in hiring, policing, and benefits determinations is already producing documented harms, and that framing all oversight as anti-competitive obscures those concrete cases.

What the experts say

Stuart Russell, a computer science professor at UC Berkeley and co-author of the standard AI textbook, has testified to Congress that current large models fail basic interpretability tests and that deployment is outpacing the science needed to verify their behavior. He has proposed licensing regimes similar to those used for pharmaceuticals.

A 2024 RAND Corporation report on frontier AI governance concluded that voluntary commitments from labs, while useful, lack enforcement mechanisms and independent auditing capacity. RAND recommended that governments build in-house technical evaluation teams before mandating industry compliance.

The Stanford Institute for Human-Centered AI’s 2025 AI Index found that reported AI incidents, ranging from deepfake fraud to autonomous-vehicle failures, rose more than 30 percent year over year.

On the China question, a 2025 Center for Security and Emerging Technology (CSET) study at Georgetown found that U.S. frontier labs retained a lead of roughly 6 to 12 months over leading Chinese models, and that export controls, not domestic safety rules, were the primary factor shaping the gap.

By the Numbers

$1 trillion: approximate combined market-cap drop across major AI-exposed U.S. tech stocks during Monday’s selloff, according to Bloomberg’s tally cited by the BBC.

6 to 12 months: estimated U.S. lead over leading Chinese frontier models, per a 2025 CSET study at Georgetown.

30%+: year-over-year increase in reported AI incidents in the Stanford AI Index 2025.

0: comprehensive federal AI safety laws passed by the U.S. Congress as of September 2026, per the Congressional Research Service.

1: Biden-era executive order on AI safety testing rescinded by Trump in early 2025, per White House records.

Hundreds: researchers who signed the 2023 Future of Life Institute letter calling for a pause on training the largest AI models.

2: U.S. senators, Markey (D-MA) and Hawley (R-MO), who have introduced separate bills this Congress requiring pre-deployment testing of frontier AI systems.