The US administration’s stance at the G20 Innovation Ministerial is unequivocal: AI progress must not be hindered by regulation, energy constraints, or major safety worries.
Held this week in Chapel Hill, North Carolina, the gathering assembles leaders from G20 nations — including India, China, Russia and numerous European states — to examine tech breakthroughs and the policy issues they raise. This year’s focus is generative AI.
Michael Kratsios, head of the Trump administration’s Office of Science and Technology Policy, told attendees that delayed policy updates hinder the ability to reap the full advantages of AI innovation.
He added that regulations drafted for older tech often mismatch fast‑moving, broadly applicable inventions, causing uncertainty for developers and policymakers, which stalls progress and lets other countries get ahead.
America is promoting the Carolina Principles — a pro‑innovation framework designed to boost AI research, streamline commercialization routes, and establish industry‑driven standards for emerging technologies, while giving major AI players a platform to speak.
Any policy accords reached by week’s end will be incorporated into the comprehensive agreement that will be signed at the G20 Leaders Summit in Miami in December.
As AI models grow more capable and autonomous, regulation becomes a pressing concern; the summit provides a window into global approaches and the boundaries being set for major tech firms.
Trump AI Adviser Warns Against a ‘DMV for AI’
David Sacks, former AI czar under Trump and current co‑chair of the president’s science and technology council, warned that adopting an FDA‑style AI regulator would be disastrous. He said such an agency would turn AI product approvals into a DMV‑like queue, delaying new models.
He contended that creating a dedicated AI regulatory body — akin to the FDA for drugs or the FAA for aviation — would be unnecessary.
While proposals for an AI‑specific regulator have surfaced before, Sacks’s comments directly target Anthropic, the prominent AI firm absent from the forum.
CEO Dario Amodei has publicly called for a governmental review process in a recent essay.
He argued that frontier AI models, comparable to aircraft, must undergo rigorous testing and auditing, and any release that fails safety standards should be halted or reversed to protect public welfare. He noted that increased AI use in research will burden existing agencies like the FDA with more review requests.
Sacks maintains that current laws and existing regulators are already equipped to address AI risks.
This isn’t the first dispute between Anthropic and the US government.
Following a collapsed Pentagon‑Anthropic agreement, the company was labeled a supply‑chain risk, but a judge subsequently halted that designation.
Later, when Anthropic introduced its Fable 5 model, the US ordered it to bar foreign nationals from using it. To comply, Anthropic withdrew the model and later republished it after receiving an informal green light from the Trump administration.
Elon Musk and Mark Zuckerberg Discuss AI
SpaceX CEO Elon Musk, who runs xAI, advocated for a light‑touch regulatory stance in a virtual address at the G20 Innovation Ministerial.
He observed that many nations over‑support established giants while neglecting nascent startups, likening the ecosystem to a forest. He urged a reversal, proposing that new inventions be presumed legal rather than illegal by default.
Musk, also the world’s richest individual and head of one of the largest tech firms, benefited from SpaceX’s recent IPO that briefly made him a trillionaire. His ventures received $38 billion in government funding last year, per a February report from The Washington Post.
Nevertheless, the US government has shown itself unable or unwilling to curb the most harmful AI outcomes. When millions of AI‑generated child sexual abuse images flooded X months ago, federal agencies took no action.
In contrast, the EU launched an investigation into X, while Malaysia, Indonesia and several other nations banned the app, and many world leaders condemned the Musk‑owned platform.
Meta CEO Mark Zuckerberg, speaking virtually, highlighted that publicly accessible AI models empower individuals to create, such as learning new software or running small enterprises.
He said the capacity to launch new ventures will be huge, creating a highly dynamic era, and that policies should facilitate easy creation and flexible evolution of existing systems.
He added that widespread diffusion enables a system of checks and balances, instead of concentrating access among a few entities or governments — a situation in which he himself would have full control.
In the past year, the Trump administration has shifted from its earlier hands‑off stance to a more interventionist one, driven by concerns that generative AI poses both threats and benefits to national security and cybersecurity.
Policy frameworks arising from forums such as the G20 Innovation Ministerial could substantially define the boundaries of power for major tech and AI firms worldwide.
Katelyn Chedraoui/CNET
