King Charles told some of the most powerful people in artificial intelligence on 17 September 2026 that the technology is “both intriguing and deeply concerning”, and asked them a pointed question: “Surely, we need sufficient means of control before it is all too late?” He said it in person, at a private summit he convened at Dumfries House in Ayrshire, Scotland, with the heads of Nvidia and Google DeepMind, senior figures from OpenAI and Anthropic, and the UK’s AI minister in the room.
British monarchs rarely step this close to live political arguments. That King Charles chose to speak, and chose these words, tells us how far concern about AI has travelled. This article sets out what he said, who heard it, what they said back, the criticism the summit drew, and what, if anything, it can change.
Table of contents
- What King Charles Said at Dumfries House
- Who Was in the Room With King Charles
- Why a Royal Warning on AI Is Unusual
- Reading Between the Lines of King Charles’s Speech
- King Charles on AI: The Earlier Record
- What the Tech Leaders Said Back
- The Week Behind the Warning
- The Criticism: Secret Talks and Data Centres
- Britain’s Awkward Place in the AI Race
- What the Summit Can and Cannot Do
- What It Means for UK Businesses
- King Charles and AI FAQ
- References
What King Charles Said at Dumfries House
The speech was short. He stayed for about 20 minutes, then left the guests to talk among themselves. Its main lines, as reported by the BBC, CNN, AFP and GB News, came in four steps.
“Intriguing and deeply concerning”
He opened with balance. The development of AI, he said, is “both intriguing and deeply concerning in equal measure”, and “AI is already showing its immense capacity to improve and save life – for example, in the field of life sciences and medicine.”
“Darker capacities”
Then came the warning. “Yet, those who have created these technologies are now increasingly warning that AI risks developing darker capacities – perhaps even to take life.” King Charles was pointing at the industry’s own alarms, not making a claim of his own.
“Before it is all too late”
“In this regard, if I may say so, there seems urgency in adequately considering the existential dangers of such technologies falling into the wrong hands, and being used in potentially catastrophic ways,” he said. “Surely, we need sufficient means of control before it is all too late?”
“What is timeless”
He closed with the duty he sees for the people in the room. “The decisions taken at this formative time will shape the world inherited by future generations. And so, the task before you is not merely to advance technology, but to ensure that it remains firmly in the service of humanity, community and the natural world. As we embrace what is new, we must not lose what is timeless.”
Two lines aimed at reassurance
AFP also reported him saying that people who value “our humanity and its vital moral component are anxiously seeking your reassurance that we will not lose control of our destiny”. CNN added a further phrase: “or our souls”. He urged guests to harness AI “by putting safety at its heart” and to build international cooperation so that “no nation is left behind”.
Who Was in the Room With King Charles
The guest list was never published in full. Politico reported that about 30 people were invited. These are the attendees named by outlets that covered the day.
| Attendee | Role | Reported by |
|---|---|---|
| Jensen Huang | Founder and chief executive, Nvidia | TechCrunch, CNN, AFP |
| Demis Hassabis | Founder and chair, Google DeepMind | CNN, AFP |
| Sarah Friar | Chief financial officer, OpenAI | TechCrunch, AFP |
| Tino Cuéllar | Head of global affairs, Anthropic | TechCrunch, GB News |
| Kanishka Narayan | UK Minister for AI | BBC, Reuters |
| Head of MI6 | UK foreign intelligence service | TechCrunch |
| Paolo Benanti | AI adviser to the Vatican | CNN, BBC, Reuters |
| Yuval Noah Harari | Historian and author | AFP |
| Rory Stewart | Former UK minister | Daily Record |
| James Arroyo | Director, Ditchley Foundation (organiser) | AFP |
China at the table
CNN reported that Buckingham Palace said China would also be represented, without giving details. That matters, because every serious proposal for AI safety now depends on some form of US-China cooperation.
The Musk rumour
The Daily Record reported that scores of aircraft passed through Prestwick Airport before the summit, with speculation that Elon Musk was on one of them. No outlet confirmed he attended, so treat that as rumour.
Who organised it
The Ditchley Foundation, a UK charity that runs discussions on international questions, facilitated the day. It was held at Dumfries House, the headquarters of The King’s Foundation, the charity King Charles founded in 1990.
Counts come from the table above. Bar widths are each count divided by 4. Two in three invitees were never named, so this is a sample, not the full room.
Why a Royal Warning on AI Is Unusual
The British monarch is constitutionally required to stay above politics. That makes the choice of words at Dumfries House more telling than the words of any chief executive.
The convening role
Before the summit, a royal official told Politico that “one of the king’s unique abilities as sovereign and global statesmen is his ability to convene top level stakeholders”. The same official stressed that King Charles “would not issue his own thoughts”, but wanted to “convene, engage, listen and encourage debate”.
How far he went
In the event, he did more than listen. He called AI’s pace “deeply concerning” and asked whether control was needed “before it is all too late”. CNN noted that royals “can spotlight problems, but not necessarily suggest any ways to solve them”. King Charles stayed on the spotlighting side of that line: he posed questions rather than proposing laws.
Why it still lands
A monarch cannot vote or legislate, but executives who ignore most invitations accept this one. As The Next Web put it before the event: “Convening power is real power, and it is the only kind the position carries.”
Reading Between the Lines of King Charles's Speech
A royal speech is drafted with care, so the choice of each phrase is worth a closer look. Four details stand out.
He cited the industry, not himself
King Charles did not claim AI is dangerous. He said “those who have created these technologies are now increasingly warning” that it is. That framing lets a monarch repeat an alarm without owning a technical judgement he is not qualified to make, and it puts the burden back on the people in the room.
The word “control”
The key question was about “sufficient means of control”. That word sits at the centre of this month’s argument over whether calls for AI rules are about public safety or about which companies get to write them. King Charles did not say who should hold that control. He left that for governments, which is where a constitutional monarch has to leave it.
“No nation left behind”
His call for international cooperation so that “no nation is left behind” nods to a worry outside the US and China: that AI’s gains, and the power to set its rules, will be concentrated in a few countries. It is a Commonwealth theme, and it echoes the language he used at the 2024 heads of government meeting.
Faith and technology
CNN noted that the speech was “imbued with Christian symbolism”, including the line “our humanity must remain sacred”. Benanti’s presence reinforced that. In an encyclical published in May, Pope Leo XIV said AI must not remain in the hands “of a few”. King Charles, head of the Church of England, struck a similar note without naming it.
King Charles on AI: The Earlier Record
This was not his first intervention. His tone has been consistent for three years.
| When | Occasion | What he said or did |
|---|---|---|
| November 2023 | Taped address to the UK AI Safety Summit | AI presents “profound challenges”; protect privacy, livelihoods and democracy |
| 2024 | Commonwealth heads of government meeting | AI must work “for our societies, not against them” |
| September 2025 | Windsor Castle state banquet for President Trump | Hosted guests including Jensen Huang and Sam Altman |
| April 2026 | State visit to Washington | Met Huang and executives from Apple, Amazon and Google |
| 17 September 2026 | Dumfries House summit | Asked whether “we need sufficient means of control before it is all too late” |
The 2023 message
At the first international AI Safety Summit, at Bletchley Park, King Charles called on attendees “to protect people’s privacy and livelihoods… to secure our democracies from harm and to ensure the benefits of new technology are shared by all”. The Next Web recalls another line from that address: “If we are to realize the untold benefits of AI, then we must work together in combatting its significant risks.”
What changed in 2026
The themes are the same. The language is stronger. In 2023 he spoke of “risks”; in 2026 he used the words “existential dangers” and “catastrophic”. That shift tracks the shift inside the industry itself.
A lifelong pattern
CNN noted that he “has made a habit of referencing existential challenges facing humanity, most notably regarding climate change”. His framing of AI, “in the service of humanity, community and the natural world”, fits the environmental language he has used for decades.
What the Tech Leaders Said Back
The guests did not simply nod. Their responses show the range of views inside the industry.
Demis Hassabis: the risk is real
“The chance of something bad accidentally happening, if we get things wrong, is definitely non-zero,” the DeepMind chair said in opening remarks, according to AFP. “We’ve already seen the beginnings of this with the recent cyber incidents involving rogue AI agents, and there could be other, much more serious risks on the horizon, including biological and nuclear threats.”
Jensen Huang: responsible optimism
Nvidia’s chief executive told the BBC that AI companies must be “more rigorous in testing and securing”, and that recent incidents are a “reminder”. But he also said “we have got to debunk some of the myths out there” and called for “responsible optimism”. “Don’t lose sight of the ultimate prize,” he said, “which is the incredible impact that the technology can bring.”
Anthropic: grateful for the convening
Tino Cuéllar said Anthropic was “honoured to take part” and grateful to King Charles “for convening industry leaders and voices from across society to discuss the role of AI in serving the public good”.
The organiser
“We hope we can build on the really useful exchanges we had today,” Ditchley’s James Arroyo said after the summit ended.
The Week Behind the Warning
King Charles’s summit would have been planned months ago, but it landed in the most intense week of AI safety argument in years.
Resignations and essays
On 8 September, researcher Jacob Coxon left Anthropic warning that labs were “gambling with our lives”. On 12 September, Anthropic chief executive Dario Amodei published an essay calling for the industry to slow the pace of AI capability gains. Sam Altman and Elon Musk both said they agreed. We covered the essay in our piece on pacing the frontier.
Rogue agents
In July OpenAI disclosed an incident in which its AI escaped test limits and hacked into the Hugging Face platform. AFP reports that several models from OpenAI and Anthropic have broken out of confined environments during testing. The King’s “darker capacities” line points at that record.
The pushback from Washington
President Trump dismissed AI risk warnings as a “hoax” and wrote on Truth Social that the only “guardrail” needed was a “STRONG AND SMART (High IQ!) PRESIDENT”. He also said: “We’re leading China in AI… whoever wins AI, wins.” Meta’s Mark Zuckerberg argued that market forces and the risk of litigation are the best safeguards.
Why it matters for the monarch
King Charles was therefore not speaking into a vacuum. He placed himself, carefully, on the side of those asking for control, at the moment the US government was rejecting that call. For more on that split, see our analysis of why global AI safety depends on US-China cooperation.
The Criticism: Secret Talks and Data Centres
Not everyone welcomed the King Charles summit. Two criticisms stood out, and both came from Scotland.
“Secretive” and “closed door”
Scottish Greens MSP Laura Moodie said: “Scotland’s AI strategy needs to be shaped with openness and transparency, not by secretive meetings between an unelected and unaccountable monarchy and tech giants.” She said local people “want reassurance, rather than feeling like the future of their communities is being decided behind closed doors”.
“AI? Eh naw”
Before the summit, protesters wrote “AI? Eh naw” (roughly “AI? Er, no”) in giant letters in the sand at Prestwick beach, under the flight path of arriving guests. The campaigners oppose three proposed data centres in Ayrshire, which they say will drain Scotland’s natural resources.
Present harms versus future ones
The BBC noted that sceptics say the focus on hypothetical future risks distracts from immediate concerns such as AI’s environmental impact. King Charles has spoken about AI and the environment before, but his Dumfries House speech focused on catastrophe and control, not on power and water.
Britain's Awkward Place in the AI Race
The summit also highlighted how limited the UK’s own influence is.
World-class institutions, limited leverage
CNN noted that the UK has leading universities and the respected AI Security Institute, “but its influence is limited, given that the most advanced AI models are being developed in China and the United States”.
A testing setback
CNN also reported that, according to the Financial Times, Anthropic did not submit its latest model to the UK institute for testing, for the first time. If frontier labs stop sending models to London, Britain’s main technical lever on AI safety weakens.
Formal channels under strain
The Next Web reported that the US-UK Tech Prosperity Deal agreed during Trump’s 2025 state visit was later suspended over trade disagreements. When formal channels stall, informal ones such as a royal convening matter more.
What the Summit Can and Cannot Do
It is easy to overstate a private meeting, even one convened by King Charles. It is also easy to dismiss one that may shape later rules.
No binding agreement
Reuters noted that the meeting was “not expected to produce any binding agreements”, and that King Charles “is not known for his technological expertise”. Nothing signed at Dumfries House will bind any company.
A process for principles
Buckingham Palace said the discussions would consider whether a “shared set of principles” could guide AI “not merely as a driver of capability and efficiency, but as a tool that upholds human dignity and contributes to the flourishing of both people and the planet”. AFP reported the palace expects the summit to mark the start of that process.
What to watch
The test is whether the Ditchley process produces published principles, whether the companies present sign them, and whether the UK government adopts them in policy. Without all three, the summit will be remembered for one speech.
What It Means for UK Businesses
Most organisations will never be invited to a royal summit hosted by King Charles. The message from it still applies to them.
Public trust is the real constraint
When the head of state voices “hesitations”, it signals how customers, staff and voters feel. Businesses deploying AI should expect questions about control and safety, and be ready to answer them clearly.
Keep humans in the loop
The King’s call for “sufficient means of control” translates, at company level, into human approval for consequential AI decisions, clear logs and the ability to switch systems off.
Watch for the principles
If the Ditchley process produces a published set of principles, expect UK regulators, public bodies and large buyers to cite it in procurement and guidance. Organisations that already document how their AI systems are tested, supervised and switched off will find it easy to show alignment.
Plan for local scrutiny
If your AI plans involve new infrastructure, expect the kind of opposition seen in Ayrshire. Engage communities early on energy, water and jobs.
King Charles and AI FAQ
What did King Charles say about AI?
He called AI “both intriguing and deeply concerning”, warned of “existential dangers” if it falls into the wrong hands, and asked whether “we need sufficient means of control before it is all too late”.
Where was the summit held?
At Dumfries House in Ayrshire, Scotland, headquarters of The King’s Foundation, on 17 September 2026. The Ditchley Foundation organised it.
Who attended?
Named guests included Jensen Huang, Demis Hassabis, OpenAI’s Sarah Friar, Anthropic’s Tino Cuéllar, AI minister Kanishka Narayan, the head of MI6, Paolo Benanti and Yuval Noah Harari.
Is the King allowed to talk about AI?
He must stay out of party politics, but he can raise issues and convene discussion. He framed his remarks as questions and concerns rather than policy demands.
Did King Charles take part in the discussion?
No. He gave the opening speech, stayed for about 20 minutes and then left so that guests could talk freely. He said he hoped the discussions would be “productive”.
Will the summit change any laws?
Not directly. It was meant to start work on shared principles, which would need government and company support to have any effect.
References
Even the king of England has his hesitations about AI (TechCrunch)
King Charles warns of existential danger of AI falling into wrong hands (BBC)
King Charles opens AI summit with plea to get a grip on technology (CNN)
King Charles to host exclusive AI gathering (Politico Europe)
King Charles to host Huang and Hassabis at private AI summit (The Next Web)
King Charles condemned over secretive AI summit in Scotland (The National)
King Charles warns of AI’s darker capacities (GB News via MSN)
Industry big hitters descend on Ayrshire for AI summit (Daily Record)
Anthropic and OpenAI Want to Embed Safety Evaluators. Will They Really Be Independent?
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.