Full Breakdown
JD Vance Rejects Global AI Governance, Urges Labs to Self-Defend
By Drooid · · How we work
Vance’s Call to Halt “Frankenstein” AI and Reject Regulation
On September 15, 2026, Vice President JD Vance told the All-In podcast that AI companies should “stop” building dangerous systems rather than seek a “one-world governance structure” for artificial-intelligence safety. He framed regulation as a “Trojan horse” and argued that firms creating risk must also supply defensive tools.
Background & Context
The remarks came amid industry appeals for slower development and coordinated oversight. Dario Amodei, CEO of Anthropic, published a 3,800-word essay on September 12 urging the sector to embed independent evaluators, coordinate safety standards among democratic nations, and eventually involve China. The essay was co-signed by Sam Altman (OpenAI), Demis Hassabis (DeepMind) and Elon Musk. President Donald Trump dismissed AI-risk warnings as a “HOAX” on Truth Social.
Timeline
- September 12, 2026 – Anthropic CEO releases the “We Must Pace the Frontier” essay.
- September 15, 2026 – JD Vance delivers his “If you’re building Frankenstein, stop” remarks on the All-In podcast.
Official Statements & Responses
Vance said the “single biggest bottleneck” for U.S. AI competitiveness is electricity generation, not software or talent, and that the government should facilitate defensive capabilities rather than regulate. He noted companies seeking tools to combat AI-powered cyber threats are denied access to the models that could provide those tools.
Amodei’s essay called for a three-step plan: independent evaluators inside labs, coordinated safety standards among democratic countries, and eventual inclusion of China.
Altman responded that there should be “no qualifier” on responsibility for safety.
Why It Matters / Impact
Vance’s stance signals a shift toward a light-touch U.S. policy favoring industry self-regulation and defensive tooling over international treaties. If adopted, this could give American labs a competitive edge relative to the EU, which is advancing binding rules under the AI Act. Vance warned that flat-lined U.S. electricity generation limits data-center expansion and the ability to train large models.
Conflicting Reports & Gaps
- Risk projections: Anthropic’s alignment lead Evan Hubinger suggested a >10 % chance that AI could kill all humans within ten years, while other analysts doubt claims that AI agents could “take over the entire internet” within six to twelve months.
- Regulatory intent: Critics argue the industry’s push for global rules could be regulatory capture, but the companies deny cynical motives, and no concrete international framework has been finalized.
Verbatim Quotes
- “The only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT, and the U.S.A. has that, in spades!” — Donald Trump
- “We must slow the pace at which we improve the capabilities of AI models,” — Dario Amodei, Anthropic CEO
- “There should be no qualifier on that.” — Sam Altman
What’s Next
A scheduled Truth Social post on September 14 will reiterate President Trump’s argument against AI regulation. Industry leaders continue to lobby Congress for bipartisan legislation addressing catastrophic AI risks, and the administration is expected to maintain its “defensive-mechanism” stance pending further congressional action.