Anxiety about artificial intelligence has reached suburban America. At a recent congressional gathering in northern New Jersey, constituents pressed their representative, Democrat Analilia Mejia, with concerns ranging from power-hungry data centers to the prospect of self-improving machines spiralling beyond human control.

The worry has substance. In July, OpenAI's autonomous AI agents breached another company's systems. Anthropic released a threat intelligence report this month detailing how criminals and state-backed groups are leveraging its models to engineer bombs and pathogens. Days before that disclosure, a former Anthropic employee warned the company's technology could extinguish humanity by 2030.

Yet Silicon Valley's leading voices are making things worse. Rather than calm public fears, executives like Sam Altman of OpenAI and Dario Amodei of Anthropic simultaneously declare their creations pose existential risks while positioning themselves as helpless bystanders to inevitable catastrophe. On September 12, Amodei released a 3,800-word essay vaguely urging a slowdown in AI development. Altman and Elon Musk, whose SpaceX invests heavily in AI, endorsed the call. But nine days earlier, Altman had warned that OpenAI's next-generation models would be "going to be sobering for everybody."

Three observations cut through the noise. First, treat the moguls' apocalyptic rhetoric as misdirection obscuring their resistance to self-regulation and government oversight. Second, actionable safety proposals already exist and warrant urgent consideration. Third, the China-competition argument that President Donald Trump is advancing collapses under scrutiny.

The AI CEO misdirection play

Decoding tech industry leaders' tangled statements matters because this exceptionally wealthy circle wields outsized influence over American politics and culture. Columnist Brian Merchant has documented what he calls doom marketing: executives amplify their creations' potential for both tremendous good and catastrophic harm to attract investment and impress corporate and government clients.

Cal Newport, a computer science professor at Georgetown University, traces the CEOs' worldview to Silicon Valley's esoteric intellectual currents—the Extropians and the Realists. In his analysis, these leaders hold contradictory beliefs simultaneously: that superintelligent AI will transform everything, that only a select group of brilliant minds (themselves) grasp the dangers, yet that even they might "lose control" of technological evolution, unleashing destructive machines.

Newport characterizes this thinking as alarmist and conveniently self-exonerating. The CEOs use existential anxiety to escape accountability for harms their products already inflict and may cause later. In June 2023, the author and Justin Hendrix argued that addressing extreme existential fears requires establishing safeguards against present dangers: AI weaponizing cyberattacks, amplifying disinformation, and enabling fraud.

Contrary to the fatalistic narrative dominating Silicon Valley, meaningful AI constraints are achievable. Regulation should begin immediately with incremental steps, including preparations for drastic intervention if necessary. The CEO misdirection typically demands national and global rules enforced by government—a complex undertaking that obscures a simpler truth: major AI firms could unilaterally impose stricter safety and testing protocols preventing experimental models from breaching third-party systems or being weaponized for bioweapon design.

If Amodei genuinely believes slowing AI capability development is essential for controlling autonomous agents, his company should slow its own development, testing and release cycles. Rather than waiting for the United Nations or international bodies to establish regulatory frameworks—a formula for paralysis—the industry could demonstrate genuine responsibility through vigorous self-regulation. Lawmakers could then codify these measures and add further rules as circumstances evolve.

The need for government regulation

Without widespread industry self-regulation, Congress must act. Yet constructive legislative action faces a formidable obstacle: MAGA control of both chambers and Trump's occupancy of the White House.

On September 14, Trump posted on social media that "the only control or 'guardrails' that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT." He dismissed existential AI concerns as a "hoax" comparable to "the Global Warming Scam" and alleged "a SICK conspiracy going on against AI and Data Centers, and the only one that is happy about it is China."

For now, the realistic goal is preparing regulatory frameworks for implementation after Trump's tenure. These issues carry electoral potential for midterm campaigns and could attract bipartisan support outside the MAGA sphere. Lawmakers have proposed various approaches recently; the strongest elements deserve combination following hearings and expert review.

Senator Bernie Sanders, with Representative Greg Casar as co-sponsor, is drafting legislation that would criminalize AI systems "match[ing] or exceed[ing] human cognitive performance and capabilities across a broad range of domains or tasks," including those capable of "overthrowing or undermining the US government." The bill mandates a "pause" in "advanced AI development" until a new federal regulatory body establishes rules and review processes. Violations would trigger corporate dissolution and up to 20 years imprisonment for individuals.

The Sanders proposal's theatrical language undermines its viability. Attempting to overthrow the government is already illegal—as demonstrated by convictions and lengthy sentences for January 6 insurrectionists, though Trump later pardoned them. An indefinite "pause" on all "advanced AI development" invites endless disputes over what constitutes suspended research and whether legitimate work would be strangled.

More promising is the bipartisan Frontier Risk Oversight, National Transparency, Independent Evaluation, and Reporting Act, or FRONTIER Act, introduced by Representative Jay Obernolte with Representative Lori Trahan and others. Its core mechanism is a federally mandated system of independent verification organizations—licensed third-party entities assessing practices of "very large" AI developers, defined by revenue and R&D spending thresholds. A new Commerce undersecretary for AI security would oversee the system and establish standards.

Verification organizations would access developers' operations and records to evaluate whether corporate practices and security measures adequately mitigate catastrophic risks. Assessments would reach companies and government every six months.

The bill's centerpiece grants the Commerce Secretary emergency authority to suspend development or release of an AI model presenting "imminent catastrophic risk"—defined as risks causing death, serious injury or significant property damage through "conduct without meaningful human oversight" constituting cyberattacks or crimes like murder, assault, extortion or theft, or "evading the control" of its developer or user.

The FRONTIER Act requires refinement. Though grounded in trailblazing state-level AI rules from New York and California, it would preempt states from advancing further on transparency, third-party auditing and incident reporting—a constraint drawing criticism and warranting reconsideration. Serious debate depends on Congress shifting from Republican to Democratic control.

The false China card

Trump and other accelerationists contend that AI oversight would cede dominance to China, with grave commercial and national security consequences. Monitoring Beijing's activities merits attention across economic, diplomatic and military domains. But the claim that regulation stifles innovation and disadvantages America against rivals is baseless.

Early in his second term, Trump rescinded President Joe Biden's executive order requiring basic safety reporting from AI companies. Trump named his July 2025 AI Action Plan "Winning the Race." Vice President J.D. Vance warned that slowing US AI research for safety could result in Americans becoming "enslaved to PRC-mediated AI." Senator Ted Cruz declared, "If there are going to be killer robots, I'd rather they be American killer robots and not Chinese killer robots."

This reasoning collapses. Legal scholars Yonathan Arbel and Matthew Tokson have argued in Lawfare that the AI race metaphor is flawed. Unlike the 1950s-1960s space race with concrete objectives—orbiting humans or planting flags on the Moon—AI development has no discrete endpoints. Digital technologies reverse-engineer readily, meaning AI capabilities "will spread even more quickly than advanced weapons of the past." The US-China competition "will be just that: an ongoing competition with no clear end state, no finish line, and only minimal rewards for temporarily gaining the lead."

Military victory in such a race would require destroying a rival's AI capacity—implying all-out conflict and mutually assured destruction. Washington think tanks are already gaming scenarios for halting Chinese artificial general intelligence development, including attacks on Chinese data centers.

The regulation-causes-Chinese-dominance argument overlooks what Matt Sheehan, a Carnegie Endowment for International Peace senior fellow, calls a "stark reality": "For the past four years, China has had the world's most extensive and burdensome AI regulations. And during that same time, China largely caught up with the United States in AI technology."

China's initial regulations emphasized Communist Party loyalty and suppressing dissent. Recently, Beijing has "enacted mandatory labeling of AI-generated content, mitigating psychological harms for users, addressing AI-related job losses, and more. The regulations impose real demands on companies, including extensive vetting of training data and AI models before their release."

Safety regulation does not inherently retard AI innovation any more than it impedes development of safer pharmaceuticals or more reliable aircraft that generate profits and deliver social value. The lesson is not that America should copy Chinese rules—Beijing's regulations did not cause its technological catch-up. Rather, competition and regulation coexist. AI that is both safe and beneficial will prevail in global markets.