Pope Leo Pushes Back After Trump Dismisses AI Doom Fears As A ‘Hoax’
Trump Calls AI Doom A ‘Hoax’ — Pope Leo Says The Risks Still Demand Attention
Pope Leo Rejects Dismissal Of AI Catastrophe Fears As ‘Fake News’
The American pope says he is not in “panic mode” over artificial intelligence, but argues that warnings from specialists cannot simply be waved away.
Pope Leo XIV has pushed back against the growing attempt to dismiss fears about advanced artificial intelligence as little more than hype.
Speaking to journalists on his flight back to Rome from France on 28 September, Leo said he sleeps at night and is not consumed by fears of an AI apocalypse. But he made an equally clear second point: warnings from specialists about catastrophic risks deserve serious attention.
“I do think that the concerns raised by many of the experts, specialists in AI, should be taken seriously,” Leo said.
He added that he did not believe those concerns were simply “fake news” created for financial or political gain.
The wording landed directly in the middle of a widening argument over how the United States should respond to increasingly capable AI.
President Donald Trump has repeatedly dismissed the idea that artificial intelligence is on course to destroy humanity. Earlier in September, he called claims that AI would take over the world and destroy humanity a “HOAX” and argued that the United States should not burden the technology with sweeping new restrictions.
Leo is taking a different position.
He is not saying catastrophe is inevitable. He is saying that uncertainty is precisely why governments, companies and civil society should keep talking about safeguards before the technology becomes harder to control.
Leo’s Position Is More Cautious Than Apocalyptic
The sharpest version of the AI debate often gets reduced to two extremes.
One side is presented as believing machines are about to wipe out humanity. The other is presented as believing that concern itself is absurd.
Leo’s comments sit between those positions.
He said he was not in “panic mode”. He also said humanity should not simply assume that nothing serious will happen.
That distinction matters.
The pope is not predicting an imminent machine rebellion. He is arguing that specialists raising concerns about loss of control, autonomous systems and future artificial general intelligence should not be dismissed without examination.
That is consistent with the position he has developed throughout his papacy.
In Pope Leo’s earlier warning about the possibility of humanity losing control of advanced AI, the central question was not whether present-day chatbots are secretly plotting against people. It was whether systems could eventually become powerful, autonomous and embedded enough that meaningful human control becomes difficult.
Leo’s latest remarks sharpen that argument.
The danger, in his telling, is not merely what AI can do today.
It is the possibility that governments and companies keep increasing capability while treating serious safety questions as obstacles to progress.
Trump Has Framed The Doomsday Argument Very Differently
Trump has taken a far more dismissive approach to claims that artificial intelligence could become an existential threat.
In a series of posts earlier this month, he said fears that AI would take over the world and destroy humanity were a “HOAX”. He also attacked calls for new guardrails and presented himself as a sufficient political safeguard against reckless development.
His broader argument is tied to competition.
The United States is in an accelerating race with China over chips, models, data centres and frontier AI capability. From Trump’s perspective, heavy restrictions risk slowing American companies while rivals continue advancing.
He reinforced that position in his 22 September speech to the United Nations, where he rejected attempts to build what he called a global scheme to control AI.
That does not mean the administration is arguing that artificial intelligence has no risks at all.
The disagreement is mainly about scale, urgency and the role of government.
Trump has resisted the idea that catastrophic scenarios justify broad regulation. Leo believes those scenarios are serious enough that political leaders and technology companies should continue investigating them before dismissing them.
The argument is therefore not simply optimism against pessimism.
It is a fight over how much uncertainty society should tolerate when the technology itself is moving quickly.
The Pope Singled Out A Contradiction In The Industry
Leo also drew attention to Nvidia chief executive Jensen Huang.
The pope referred to new efforts to build stronger technical guardrails into AI systems, then questioned how that could sit alongside arguments against government limits.
His point was straightforward.
If companies themselves are building safety systems because advanced AI may behave unpredictably, it becomes harder to argue that safety concerns are entirely manufactured.
The existence of guardrails does not prove that catastrophic AI is coming.
It proves that developers already recognise a control problem at smaller scales.
Modern AI systems can hallucinate, misuse tools, follow badly specified objectives and behave differently outside controlled evaluations. Most of those failures are nowhere near an extinction-level event. But the safety debate is about what happens as systems become more capable and are given more authority.
That is especially important with AI agents capable of planning and carrying out multi-step tasks.
A passive chatbot responds.
An autonomous system may be allowed to browse, write code, operate software, move information or make decisions over a long sequence of steps.
Capability and risk can therefore grow together.
Today’s Systems Are Powerful, But They Are Not Superintelligence
The phrase “AI doom” can make the debate sound more settled than it is.
It is not.
There is no public evidence that current AI systems are artificial superintelligence. There is no scientifically defensible countdown showing that humanity will lose control of machines in a particular year. There is no proof that extinction is the inevitable endpoint of continued AI development.
Current systems remain unreliable in important ways.
They can produce fluent nonsense. They can misunderstand instructions. They can fail on tasks that humans find simple. They often require substantial infrastructure, supervision and access before they can do meaningful work in the world.
That is why the difference between today’s models and genuine artificial superintelligence matters.
The strongest safety argument is not that catastrophe has already arrived.
It is that uncertainty increases as capability increases.
A system does not need consciousness, anger or a desire to overthrow humanity to become dangerous. It only needs enough competence, autonomy and access to produce consequences its operators cannot easily reverse.
That is the problem Leo wants discussed before it becomes urgent.
The Debate Is Moving Into The White House
The timing of Leo’s intervention is significant.
Trump and House Speaker Mike Johnson are meeting senior technology executives in Washington on 29 September as the United States confronts growing pressure to define its approach to AI oversight.
Executives expected at the discussions include figures from Nvidia, Meta, Anthropic, OpenAI and Google.
The central question is familiar.
How can the United States keep pushing forward without creating risks it cannot later contain?
Johnson has argued against both a moratorium and heavy-handed regulation, while saying that government still needs to find a balance between innovation and oversight.
That position reflects the wider political difficulty.
Move too slowly and dangerous systems may be deployed before institutions understand them.
Move too aggressively and policymakers risk freezing useful technology, strengthening incumbents or handing strategic advantage to overseas competitors.
There is no simple formula that removes that tension.
AI Leaders Are Not Speaking With One Voice
The technology industry itself is divided.
Some executives have warned that increasingly powerful systems could create catastrophic risks and have called for stronger coordination between laboratories and governments.
Others have argued that public fears are exaggerated and that technical safeguards, internal testing and market incentives can address most problems without aggressive state intervention.
Those disagreements matter because “the AI industry” is often discussed as though it has one position.
It does not.
Even within the same company, researchers focused on capability, safety, deployment and commercial growth can face different incentives.
Recent incidents and cancelled releases have also kept safety questions alive. Taylor Tailored has examined how OpenAI scrapped a planned GPT-6.1 Astra release after safety testing raised concerns.
That does not prove that frontier AI is becoming uncontrollable.
It does show why serious developers test for failure before deployment.
Safety work exists because systems can behave in ways their creators do not want.
Leo Wants The Vatican Inside The AI Conversation
Leo has made artificial intelligence one of the defining technological issues of his papacy.
The Vatican has created its own commission on AI, both to examine how the Holy See uses the technology and to explore broader risks and benefits.
Leo has also called for political leaders, technology executives, social organisations and researchers to keep talking to one another.
His language is moral rather than technical, but the underlying question is practical.
Who remains responsible when machines make more decisions?
The concern reaches well beyond a hypothetical superintelligence.
AI is already moving into education, employment, warfare, healthcare, finance, government and media. Systems are being trusted with recommendations and decisions that can affect real people long before any machine becomes smarter than humanity.
Leo’s wider argument is that human judgement cannot be treated as an inconvenience that should automatically be engineered away.
The Real Divide Is About What Happens Before We Know
Trump and Leo are both reacting to the same uncertainty.
They are drawing different conclusions from it.
Trump’s position is that apocalyptic claims are being overstated and that excessive regulation could damage American leadership in a technology with enormous economic and strategic value.
Leo’s position is that uncertainty is not a reason to stop asking hard questions.
It is a reason to keep asking them.
Neither position changes the underlying facts.
Artificial intelligence is improving rapidly. Current systems are not proven superintelligence. Catastrophic loss of control remains a future risk rather than an established present reality. Governments are still struggling to decide how much oversight is proportionate.
The disagreement is over what should happen while nobody can confidently predict where the technology ends.
Leo’s answer is cautious.
Do not panic.
Do not pretend certainty.
And do not wait for a crisis before deciding who should remain in control.
Sources
Reuters — Pope Leo says concerns about AI doom scenarios are not ‘fake news’ — Supports Leo’s 28 September remarks, his comments on Nvidia and the timing of the White House AI meeting.
Associated Press — Trump calls AI risks a ‘hoax,’ says there is a ‘SICK conspiracy’ against AI and data centers — Supports Trump’s September statements dismissing AI doomsday fears and his opposition to new guardrails.
Vatican News — Pope Leo XIV: AI must serve human dignity and strengthen peace — Supports Leo’s broader position that AI development should remain subordinate to human dignity and responsibility.
Next Reads
Pope Leo Warns Of AI Apocalypse Risk And Humanity Losing Control — The closest background piece on Leo’s wider AI warnings and his loss-of-control argument.
What Is Artificial Superintelligence — And How Close Are We To Creating It? — Explains the distinction between today’s AI and the hypothetical systems at the centre of existential-risk fears.
OpenAI Scraps GPT-6.1 Astra Release After Safety Tests — What Went Wrong? — A current example of how frontier AI developers use safety testing before deployment.
