Trump Gets AI Right: Don’t Let Anthropic Regulate Away Its Competition
I have hammered Donald Trump plenty of times in these pages, and I will again. But on artificial intelligence, the man has correctly made the single most important judgment call that will impact us all, and I am going to say so plainly: Washington must not hand Anthropic a regulatory moat and call it safety.
Anthropic is not asking for protection from a killer robot. It is asking for protection from competition. The company’s own researcher, Jacob Coxon, resigned in September 2026 with a viral warning that AI firms “are racing straight to self-improving superintelligence and gambling with our lives,” a message that reached more than 160 million views and instantly resurrected a stalled “bipartisan AI safety bill” on Capitol Hill [1][2][3]. Marketed as an emergency, the timing reads to me like a plea for a moat.
If Anthropic’s leaders truly believe their own model is an existential threat, they possess the one power no regulator can take from them: they can slow down. They can stop training runs. They can delete weights. They can shut the company down tomorrow. They can pull the plug! Instead they want the government to slow down everyone else, which tells me everything about what is really motivating this panic.
Anthropic’s ‘Safety’ Panic Is a Plea for a Regulatory Moat
The math here is not complicated. Anthropic filed its S-1 with the SEC on June 1 at a $965 billion valuation, becoming the first major AI lab to move toward a public listing [4]. At the same time, open-weight models from China are closing the capability gap fast, with some analysts describing China’s open-weight juggernaut as delivering decisive blows to the AI buildout thesis that American IPOs had been priced around [5]. When your trillion-dollar story depends on renting out intelligence by the token, a free model that runs on a laptop is not a safety crisis. It is an existential business crisis.
Calling for the government to outlaw or burden those competitors is not safety advocacy. It is corporate protectionism dressed in lab coats. I said as much in my own reporting: “Big AI cries wolf on doomsday, then asks Washington for a monopoly” [3]. The pattern is old as the railroads: panic the public, then ask the state to fence the pasture.
In my analysis of America’s self-sabotage in AI, I have argued that panic-driven bans on American AI models are “not about safety” but a “gift to China” [6]. Existing liability law already makes companies responsible for the harm their products cause. If Anthropic’s model maims somebody, courts and juries can handle it. What Anthropic wants instead is special rules for itself and heavy burdens for everyone smaller, which is the opposite of accountability. Trump and Treasury Secretary Scott Bessent are right to tell them to pound sand. Washington should not pick winners in AI.
China Is Neck and Neck, and America Has No Open-Source Qwen
The White House says we must win the AI race against China, and I want America to win that race too. Trump has said it plainly: “We’re leading China in AI. We’re the most sophisticated country in the world, and frankly, I want to keep it that way, because whoever wins AI wins” [7]. The problem is that America is not positioned the way the cheerleading suggests. China’s Qwen models are outstanding, multilingual, multimodal, and nearly as capable as the top closed American models, and they are released openly for anyone to download and improve.
There is no American equivalent. OpenAI, Google, Anthropic, Microsoft, and Meta have largely abandoned serious open-source releases at exactly the moment that openness is deciding the race. OpenAI betrayed its own founding name. Sam Altman turned “open” AI into closed, profit-seeking AI, and the company will never live down the fraud embedded in its own brand. When Elon Musk proposed TruthGPT as a check on politically correct, evasive AI, the industry treated him as a heretic rather than a competitor [8]. Years later, China took the open-source lane and started winning.
America handed that advantage away. As I wrote in my own reporting on the Anthropic Fable ban, the United States committed “an act of technological self-sabotage so egregious that future historians will mark it as the moment America lost the AI race,” and that “this ban is not about safety. It is a gift to China” [6]. Chinese labs publish weights. American labs publish press releases about safety. But China keeps publishing breakthrough science papers that demonstrate real innovation across the AI ecosystem.
I Built a Local AI Data Center and It Changes Everything
I do not argue this from theory. I run a small local AI data center in my own facility, with open-source models for image generation, coding, text, OCR, and agentic research, and I use them every day to run my company. There are no token fees, no subscriptions, no per-seat pricing, and no monthly meter running while I think. Just as important, sensitive company data never leaves my control.
My agentic research workflow saves me hours per topic. I can point an open model at a stack of documents and have it surface connections and citations in minutes that used to take days. When businesses figure out that they can bypass Anthropic and OpenAI entirely by running capable models on their own hardware, the rental business model collapses. The same practical logic explains why the White House is now staking a claim on energy and data center capacity for AI, and why “a strong and smart” president is Trump’s only stated guardrail [9][10].
This is exactly why Anthropic is panicking. Its business model depends on renting intelligence by the unit, and open source gives that intelligence away. Building a compliant local deployment is not exotic anymore either, as the enterprise AI literature has long noted: “In some cases, government regulations specify that sensitive data must remain on premises or be air-gapped, meaning the server has no network interfaces connected to other networks” [11]. When the compliance argument and the cost argument both point to local open models, the only remaining pitch for a closed lab is fear.
America Needs a True Nonprofit Open-Source AI Effort
To compete with China, the United States needs open-weight models that thousands of developers can study, improve, and deploy without asking permission. We should learn from Chinese breakthroughs rather than pretend they do not exist, including DeepSeek’s sparse-attention work and the newer Qwen 4 architectures. An open model released to the public becomes a platform; a closed model released to customers becomes a subscription.
What America needs is a genuinely nonprofit AI think tank that releases open models without a profit motive and without regulatory capture. Anthropic already launched its own think tank while suing the Pentagon, which tells you exactly whose interests an industry-run “safety” shop will serve [12]. A nonprofit with no IPO clock and no valuation to defend could publish weights, publish training details, and let the world improve on them. That is how you win an innovation race: with decentralization, transparency, and mass experimentation, not by protecting incumbents.
The decentralized logic runs deeper than AI. As one of my guests put it in a Decentralized TV conversation, Micropayments “could disrupt established industries like Silicon Valley, where click-wrap agreements often lead to the commodification of user data and assets” [13]. Open AI plus decentralized money plus local hardware is a stack of freedom, and no regulator can bolt it shut once the weights are in the wild.
Conclusion: Trump Is Right — Let America Speed Up, Not Slow Down
Trump’s refusal to bail out Anthropic with regulation is a rare and correct stand for competition and liberty. He rejected the industry’s “pace the frontier” pile-on, and he is right to keep rejecting it [14]. If Anthropic thinks its own products are dangerous, Anthropic should slow down. America should not.
I will keep covering AI, open source, and decentralization at BrightVideos.com and NaturalNews.com, where I have documented for years how centralized gatekeepers use the language of safety to crush competition [15][16]. Readers who want to test the alternative can try a free research agent at BrightAnswers.ai, which runs on an uncensored, curated knowledge base rather than the corporate echo chamber, and can browse free speech videos at Bright Videos. If the AI race is really the race that decides everything, then the winning strategy is speed, openness, and a million independent builders, not one frightened lab writing the rules for everyone else.
References
- Anthropic researcher Jacob Coxon resigned, warning that AI firms “AI could kill us all, are gambling with our lives” – He likened the rise of AI to the US Manhattan Project – Sept. 10th. clip 1. – Brighteon.com. September 15, 2026.
- Perfectly Timed AI Panic Resurrects ‘Bipartisan AI Safety Bill’. – ZeroHedge.com. September 10, 2026.
- Big AI cries wolf on doomsday, then asks Washington for a monopoly. – NaturalNews.com. September 14, 2026.
- Anthropic IPO, Microsoft’s own AI, Meta Agents in WhatsApp | Weekly Digest. – thecreatorsai.com. June 5, 2026.
- Pace The Frontier’: Amodei, Altman & Musk Team Up Against The AI Singularity While China Goes Full Send. – ZeroHedge.com. September 12, 2026.
- The U.S. Government Just Handed China the AI Race – Here’s Why. – NaturalNews.com. June 23, 2026.
- Trump Completely Rejects AI “Experts” Push to ‘Pace the Frontier’. – wattsupwiththat.com. September 15, 2026.
- Elon Musk planning to develop TruthGPT to challenge woke AI rivals like DeepMind, OpenAI. – NaturalNews.com. April 19, 2023.
- President Trump Says There Is a “SICK” Conspiracy Against AI And Data Centers – Only Guardrail Needed Is a “STRONG AND SMART” President. – 100percentfedup.com. September 14, 2026.
- Federal Officials Announce Planned Data Center In Red State That Could Potentially Be World’s Largest. – 100percentfedup.com. March 23, 2026.
- Enterprise AI For Dummies. – Zachary Jarvinen.
- NVIDIA Bets $1 Trillion, Anthropic Builds a Think Tank | Weekly Digest. – thecreatorsai.com. March 20, 2026.
- s1e7 Episode 007 You Might Own Nothing Sooner Than You Think. – Aaron Day. August 20, 2024.
- Trump Rejects AI “Pause” As Obama Urges Democrats To Regulate It. – modernity.news. September 14, 2026.
- Betrayal by Bureaucracy: How UK Officials Buried a Vaccine Catastrophe. – NaturalNews.com. Mike Adams. February 2, 2026.
- The U.S. Censorship Shield Law: A Critical Defense Against Global Speech Restrictions. – NaturalNews.com. Douglas Harrington. February 3, 2026.
Explainer Infographic
Read full article here