Trump Says AI Risk Is a Hoax. Anthropic and Its Rivals Disagree.
The US president claims his IQ alone is enough to keep AI safe. The companies actually building the technology have been calling for a coordinated global pause.
Key points
- Donald Trump posted on Truth Social that fears about artificial intelligence risk are a hoax invented by conspiracy theorists.
- Anthropic and rival AI labs have publicly called for a coordinated global pause to give the industry time to address safety concerns.
- Trump claimed that a president with a "very high IQ" is sufficient protection against any danger posed by AI systems.
- No independent safety body or peer-reviewed evidence supports the claim that AI risk concerns are fabricated.
Donald Trump posted on Truth Social this week that concern about artificial intelligence is a hoax dreamed up by conspiracy theorists. The claim was reported by The Guardian.
It would be easier to dismiss if the people raising those concerns were fringe voices. They aren't.
Anthropic, the AI safety company behind the Claude family of AI assistants, has joined competitors in calling for a coordinated global pause while the industry works out how to reduce the risks. The companies building these systems are the ones asking for breathing room. That's not the behaviour of fantasists.
This is the second Truth Social post on AI safety we've covered this week: our 14 September story "Trump calls scrutiny of AI a 'SICK conspiracy'" found the president warning unnamed critics of consequences without citing a single concrete example.
What did Trump actually claim?
His argument, stated plainly: a president with a very high IQ can out-think any AI system and tell it where to draw the line. According to Trump, the machines would then reprogram themselves accordingly.
That's not how AI systems work. Large language models, the technology behind tools like ChatGPT and Claude, don't contain a default off switch keyed to the US presidency. They're trained on enormous datasets by teams of engineers and don't rewrite their own code in response to verbal instructions from political leaders.
The gap between what Trump described and how these systems actually function isn't a matter of opinion. It's engineering.
Why does this matter for ordinary people?
Policy shapes what safety rules AI companies must follow, and right now the US federal government sets the tone for much of the world. If the administration treats documented safety concerns as conspiracy theory, the practical result is less pressure on companies to accept independent oversight.
For patients who'll increasingly encounter AI in diagnosis and drug development, and for workers whose jobs AI is reshaping, the question of who is watching the watchers is not abstract.
Anthropic's researchers, and those at competing labs, have spent years publishing detailed accounts of where current AI systems fail or behave in ways their creators didn't intend. Calling that body of work a hoax doesn't make the failure modes disappear.
Here's the honest read: this is a politician telling a technically demanding story in the simplest possible way, and the simplification is so severe it becomes fiction. The safety debate is real. The timelines are contested. The risks are genuinely uncertain, and uncertain isn't the same as non-existent.
Common questions
Do AI companies actually think their own technology is dangerous?
Some of the largest ones say yes, at least in certain scenarios. Anthropic, which builds Claude, was founded specifically around making AI safer and has called publicly for managed slowdowns alongside competitors.
Can a government leader simply instruct an AI to behave safely?
No. AI systems are shaped by training data and engineering choices made by developers. A declaration from any political leader doesn't alter how a deployed model behaves.
