Trump calls the AI extinction fear a "hoax"; a Republican senator answers "Come on, it's not a hoax"

Senators are expected to debate a mandated shutdown switch for the most advanced AI models today, after Dario Amodei, Sam Altman and Elon Musk asked developers to slow down. Fifteen reports split on whether the risk is real, and on who the industry leaders are.

Against AI Doomerism as published by National Review · seen at National Review

United States Published 16 Sep 2026, 19:56 UTC United States

The president's word. The Washington Examiner reports Trump called the doomsday concerns a "hoax" while calling AI and the data centres that power it the "Greatest Economic Development Engine in History". CNBC writes he "has angrily pushed back on the calls from industry and lawmakers for an AI slowdown".

The answer from his own side. The Hill quotes Sen. Thom Tillis (R-N.C.): "Come on, it's not a hoax. It's a real thing." Sen. Bernie Sanders (I-Vt.) told MS NOW it is "embarrassing… how ignorant this president is about the dangers of AI".

Where the accounts split.

Outlet	What it says the argument is
Washington Examiner	industry leaders asking to be slowed down threw Congress a "monkey wrench"
Fox News	"doomer"-driven rules would hand China an edge and entrench an "OpenAI-Anthropic duopoly"
National Review	"Against AI Doomerism" — "We can't afford to fall behind"
The Atlantic	"Destroying Humanity Is Against the Law"
The Hill	a POLITICO survey: 63% of US adults think AI could someday destroy mankind
The New York Times	a Times/Siena poll: voters still uncertain, AI "not yet a national political issue"

Not settled. Neither poll report states a sample, field dates or question wording. The Washington Examiner writes flat, sourced to nobody, that an OpenAI model "went rogue during internal testing and hacked into the company Hugging Face"; no other report we hold carries that episode. And nothing we hold says what today's Senate debate produces.

What this is about. Congress is deciding whether to force the American companies building the most advanced AI models to fit them with a way to switch them off — while the president says the danger such a law would answer is not real.

How it got here. Anthropic, OpenAI and Elon Musk's xAI are the US firms building frontier models, and Washington has been split for years over whether to regulate them at all; the Washington Examiner describes lawmakers as "largely divided", with Republicans and Democrats spread across the spectrum. This month the argument turned round. A former Anthropic researcher warned publicly that AI models could kill off humans "by the end of the decade" — a warning the Examiner says "sent shockwaves through the U.S. public". On Saturday, Anthropic's chief executive Dario Amodei published an essay urging tech leaders to slow the pace of development, and on CBS News the next day said a kill switch "could be a good idea". Amodei, OpenAI's Sam Altman and Musk then jointly called on developers to pump the brakes — an industry asking to be restrained, which the Examiner calls the "monkey wrench" thrown at Capitol Hill this week.

Who is in it. Rep. Ted Lieu, a California Democrat, and Rep. Nathaniel Moran, a Texas Republican endorsed by Trump for re-election, have introduced the AI Kill Switch Act, which would require developers to build a shutdown mechanism. Trump takes what the Examiner calls a "traditionally conservative, anti-regulation stance". Sanders and Steve Bannon, at opposite ends of American politics, have both lined up behind caution.

What is at stake. A mandate would require something that does not yet reliably exist. Anthropic said in an April court filing it had no kill switch; OpenAI says it is "building toward" autonomous shutdown; the head of Berkeley's Center for Human-Compatible AI told Scientific American such switches are "more of an aspiration than a reality". Against that, a former Pentagon AI official told Fox News regulation would hand China an edge.

Where it stands. Senators are expected to debate the idea on Wednesday. No bill has passed.

What is being argued about

Whether AI poses an existential risk at all, or whether saying so is a hoax
A bipartisan bill requiring AI developers to build a shutdown mechanism
What the public thinks — and whether either poll shows its method
Whether the industry's leaders are petitioners for regulation or a cartel seeking one

One argument, seven accounts of what it isSide by side

Outlet	What it says the argument is	Whose words
Washington Examiner	Trump called the doomsday concerns a "hoax"; a Trump-endorsed Republican co-sponsors the shutdown bill	Trump, Rep. Nathaniel Moran
The Hill	"Come on, it's not a hoax. It's a real thing."	Sen. Thom Tillis (R-N.C.)
CNBC	Trump "has angrily pushed back on the calls from industry and lawmakers for an AI slowdown"	the outlet's own
Fox News	"doomer"-driven regulation could hand China an edge and entrench an "OpenAI-Anthropic duopoly"	Will Thibeau, former Pentagon AI official
National Review	"Against AI Doomerism" — "We can't afford to fall behind"	the outlet's own, in the first person plural
The Atlantic	"Destroying Humanity Is Against the Law" — don't wait for a mass-extinction event to hold developers accountable	the outlet's own
The New York Times	voters are still uncertain, including on which party would do better; AI is not yet a national political issue	a Times/Siena poll

Each cell is what that report puts at the centre of the dispute, in its own words where it used them.

Sources

Washington Examiner

The Hill

CNBC

Fox News

National Review

The Atlantic

The New York Times

Five days from an essay to a Senate debateTimeline

Sat 12 Sep— no time published Dario Amodei, Anthropic publishes an essay urging tech leaders to slow the pace of AI developmentdated by The Hill
Sun 13 Sep— no time published Dario Amodei, Anthropic tells CBS News an AI kill switch "could be a good idea"dated by The Hill
Mon 14 Sep— no time published Elon Musk, xAI calls for greater AI regulation, proposing peer review before companies release modelsdated by the Washington Examiner
Tue 15 Sep— no time published Jack Clark, Anthropic co-founder tells the BBC it would be hazardous to leave the industry "totally unregulated"dated by the Washington Examiner
Tue 15 Sep— no time published Sen. Thom Tillis (R-N.C.) tells reporters at the Capitol "Come on, it's not a hoax"dated by The Hill
Wed 16 Sep13:40 published on the report The Hill reports a POLITICO survey released that morning 63% of US adults
Wed 16 Sep16:29 published on the report Washington Examiner reports senators are expected to debate a mandated kill switch that day

Times are the publication times printed on the reports, in UTC. Events described inside a report carry no published time.

Sources

The Hill

Washington Examiner

The Hill

Claims and verdicts

unverified 63 percent of US adults believe AI could someday destroy humanity.

Only The Hill carries the figure, attributing it to a POLITICO survey, and gives no sample, dates or method. The New York Times cites a separate Times/Siena poll showing voters still uncertain about AI, which neither confirms nor contradicts the number.

Checked against 2 sources:

The Hill

The New York Times

true President Trump called concerns that AI poses an existential danger a "hoax".

The Washington Examiner quotes Trump calling the doomsday concerns a "hoax" while praising AI data centres, and The Hill carries Sen. Thom Tillis answering "Come on, it's not a hoax"; CNBC separately reports Trump pushing back angrily on calls for a slowdown. No report we hold disputes that he used the word.

Checked against 3 sources:

Washington Examiner

The Hill

CNBC

true Rep. Ted Lieu and Rep. Nathaniel Moran have introduced a bipartisan bill requiring AI developers to build a shutdown mechanism.

The Washington Examiner describes the AI Kill Switch Act co-sponsored by Moran and Lieu and lists what it would require, and The Hill independently refers to the kill switch Lieu and Moran have proposed. Neither report is contradicted by any other we hold.

Checked against 2 sources:

Washington Examiner

The Hill

Who wrote it

M Molly Parks Washington Examiner

What the bias passes found in their reporting

Accused response minimised Saudi Arabia's denial gets a four-word clause and no quote×1

Asserted versus attributed the administration's description of the museums is quoted and left standing×1

Contested incident stated flat an unverified rogue-model episode is written as settled fact, attributed to nobody×1

Denial placement The Saudi denial gets one subordinate clause, far down the report, after long quotes from Strada.×1

Claims we checked that rest on their reporting

1 misleading2 disputed5 unverified5 true

From 166 reports we hold under this byline, at Washington Examiner. Two independent readings, each made without the masthead, a quote attached to every tag. A record, never a rating. Their full record →

C Charles Creitz Fox News

What the bias passes found in their reporting

A number, with no people named The only report that gives a figure counts the journalists whose communications were swept up. It names none of them.×1

Measure chosen The report gives a count of journalists whose communications were swept up, which widens the story beyond lawmakers.×1

Single source supplies the frame one former official's warning is the whole report, and his label for the other side rides in the headline×1

What one side is called the safety side is labelled in the headline; the other side is not labelled×1

Claims we checked that rest on their reporting

1 unverified

From 69 reports we hold under this byline, at Fox News. Two independent readings, each made without the masthead, a quote attached to every tag. A record, never a rating. Their full record →

M Mira Bhakta The Hill

What the bias passes found in their reporting

Asserted versus attributed racist placed in quotation marks and credited to the students×1

Asserted vs attributed stated flat that the process bypasses Congress×1

Attributed 'racist' placed in quotation marks as the students' word×1

Attribution puts the demand for an investigation on two named lawmakers×1

Claims we checked that rest on their reporting

2 disputed5 unverified5 true

From 246 reports we hold under this byline, at The Hill. Two independent readings, each made without the masthead, a quote attached to every tag. A record, never a rating. Their full record →

S Sarah Davis The Hill · 2 reports on this story

What the bias passes found in their reporting

Asserted not attributed States the purpose of the payments flat, in the administration's terms.×1

Asserted versus attributed the same two days listed without a characterising adjective on any speech×1

Asserted vs attributed states the purpose flat, in the administration's terms, with no attribution×1

Attributed characterisation Carries Trump's description of the fund. The description is attributed but not set against any account of what the fund does.×1

Claims we checked that rest on their reporting

1 disputed14 unverified14 true

From 223 reports we hold under this byline, at The Hill. Two independent readings, each made without the masthead, a quote attached to every tag. A record, never a rating. Their full record →

A Akayla Gardner MSNBC

What the bias passes found in their reporting

No actor named Written by the person it fell on; the president bans in one sentence, then the badge is taken away with nobody named as taking it×1

Who is named as acting the president acts in the first sentence; the badge is taken with no actor named×1

Claims we checked that rest on their reporting

1 true

From 16 reports we hold under this byline, at MSNBC. Two independent readings, each made without the masthead, a quote attached to every tag. A record, never a rating. Their full record →

T Tara Suter The Hill

What the bias passes found in their reporting

Allegation placed with the woman The allegation is attributed to the woman herself, not left without a source.×1

Asymmetric labelling the Speaker carries party and state in the first clause×1

Headline beyond body one suspected case becomes an 'outbreak' in the headline×1

Headline built from one side's words the critic's adjectives carry the headline unquoted-in-body form×1

Claims we checked that rest on their reporting

1 misleading2 unverified1 true

From 205 reports we hold under this byline, at The Hill. Two independent readings, each made without the masthead, a quote attached to every tag. A record, never a rating. Their full record →

J Juliette Kayyem The Atlantic

No independent reading has coded a report under this byline yet. We do not infer a journalist's views from the outlet that employs them.

From 1 report we hold under this byline, at The Atlantic. Two independent readings, each made without the masthead, a quote attached to every tag. A record, never a rating. Their full record →

K Kim Wehle The Hill

No independent reading has coded a report under this byline yet. We do not infer a journalist's views from the outlet that employs them.

Claims we checked that rest on their reporting

1 true

From 4 reports we hold under this byline, at The Hill. Two independent readings, each made without the masthead, a quote attached to every tag. A record, never a rating. Their full record →

N Nate Cohn The New York Times

What the bias passes found in their reporting

Hedged headline the poll's own publisher hedges its conclusion×1

Speculative headline The headline argues that Democrats have a strategy, which goes beyond reporting the poll numbers.×1

Claims we checked that rest on their reporting

1 unverified

From 9 reports we hold under this byline, at The New York Times. Two independent readings, each made without the masthead, a quote attached to every tag. A record, never a rating. Their full record →

E Eli Tan The New York Times

What the bias passes found in their reporting

Inference hedged twice a motive is read off a post and marked as reading, not reporting×1

From 15 reports we hold under this byline, at The New York Times. Two independent readings, each made without the masthead, a quote attached to every tag. A record, never a rating. Their full record →

T Toluse Olorunnipa The Atlantic

No independent reading has coded a report under this byline yet. We do not infer a journalist's views from the outlet that employs them.

From 4 reports we hold under this byline, at The Atlantic. Two independent readings, each made without the masthead, a quote attached to every tag. A record, never a rating. Their full record →

The bylines on the reports this story is built from. Hover one for what our passes have measured on their reporting. · 3 reports we hold carry no byline — we do not invent one.

Coverage, outlet by outlet

Washington Examiner — independent

How a mandated ‘kill switch’ could help regulate AI

The fullest account: what a kill switch is, what the Lieu–Moran bill would require, and what each company says it has — and the only report asserting, with no source attached, that an OpenAI model went rogue in testing.

Senators are expected to debate the concept of an artificial intelligence “kill switch” on Wednesday, as Capitol Hill reckons with calls from AI industry leaders to regulate its most advanced models. Lawmakers in Washington, D.C., who have been largely divided on the issue of AI regulation, were thrown a monkey wrench this week when industry leaders like Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and SpaceXAI CEO Elon Musk joined forces to call on developers to pump the brakes on frontier AI model progress. Recommended Stories How a mandated ‘kill switch’ could help regulate AI Sam Alt

Hedged or stated flat: stated flat — an unverified incident is asserted as fact with no source attached — “The concept of a “kill switch” gained momentum in the public policy world after an advanced OpenAI model went rogue during internal testing and hacked into the company Hugging Face.”

Contested incident stated flat: an unverified rogue-model episode is written as settled fact, attributed to nobody — “The concept of a “kill switch” gained momentum in the public policy world after an advanced OpenAI model went rogue during internal testing and hacked into the company Hugging Face.”

Read it at Washington Examiner → https://www.washingtonexaminer.com/policy/technology/4729292/how-mandat…

CNBC — independent

Pentagon CTO says U.S. government shouldn't take stakes in tech giants, questions adding AI rules

The Pentagon's chief technology officer questions adding AI rules and says the government should not take stakes in tech giants.

President Donald Trump has angrily pushed back on the calls from industry and lawmakers for an AI slowdown or greater regulation.

Read it at CNBC → https://www.cnbc.com/2026/09/16/emil-michael-ai-regulation-trump.html

CNBC — independent

Anthropic, OpenAI proposed new 'neutral' AI watchdogs. Why you should worry about the idea

Examines what Anthropic and OpenAI actually proposed — embedded "neutral" evaluators — and why the idea has problems.

Anthropic and OpenAI are proposing embedded AI evaluators to help manage risk of models causing catastrophic harm to society, but the idea has some issues.

Read it at CNBC → https://www.cnbc.com/2026/09/16/anthropic-open-ai-model-safety-risks.ht…

Fox News — independent

AI 'doomers' risk kneecapping US vs China, but ex-Pentagon vet says both extremes miss the mark

The only report making the China argument, and the only one naming a risk that regulation entrenches an "OpenAI-Anthropic duopoly".

"Doomer"-driven AI regulation could kneecap the U.S. against China while helping tech giants lock in their lead over smaller startups, a former Pentagon AI official who also worked for AI company Palantir Technologies told Fox News Digital. The warning follows the resignation of an Anthropic researcher who said increasingly powerful AI could threaten humanity, while President Donald Trump called fears of rogue AI a "hoax" and compared those advancing such warnings to climate-change alarmists whose predictions, he argued, haven't come true. "The notion that I think the president and I share is

What one side is called: the safety side is labelled in the headline; the other side is not labelled — “AI 'doomers' risk kneecapping US vs China, but ex-Pentagon vet says both extremes miss the mark”

Single source supplies the frame: one former official's warning is the whole report, and his label for the other side rides in the headline — “Will Thibeau, a former Pentagon AI official, warns that "doomer"-driven AI regulation could hand China an edge and entrench an OpenAI-Anthropic duopoly.”

Read it at Fox News → https://www.foxnews.com/politics/ai-doomers-risk-kneecapping-us-vs-chin…

The Hill — independent

‘You can’t trust anything’ from Anthropic, OpenAI leaders, lawyer says

A lawyer suing OpenAI and Anthropic says "you can't trust anything" their leaders say; the report names his own suits in the same sentence.

Attorney Jay Edelson slammed the heads of the leading artificial intelligence companies Tuesday, declaring “you can’t trust anything” they say. Appearing on News Nation with Leland Vittert, Edelson – who is currently leading lawsuits against OpenAI, Anthropic and other major technology companies – was asked if he had ever seen leaders within an industry abdicate…

Interest disclosed inside the sentence: the lawyer's own suits against the companies he attacks are named in the same breath as his attack — “Edelson – who is currently leading lawsuits against OpenAI, Anthropic and other major technology companies – was asked if he had ever seen leaders within an industry abdicate”

Read it at The Hill → https://thehill.com/policy/technology/6092873-edelson-slams-ai-leaders/

The Hill — independent

Senate Republican on Trump claim AI threat is ‘hoax’: ‘Come on’

The only report in which a Republican senator answers the president's word directly: "Come on, it's not a hoax."

Sen. Thom Tillis (R-N.C.) on Tuesday criticized President Trump’s dismissal of rising concerns about the dangers of artificial intelligence technology. “Come on, it’s not a hoax. It’s a real thing,” Tillis, who is not running for reelection, told reporters at the Capitol. “This is something that I’ve spent a lot of my professional career in,…

Who is named as acting: the senator criticises; the president's position is a noun, not an act — “Sen. Thom Tillis (R-N.C.) on Tuesday criticized President Trump’s dismissal of rising concerns about the dangers of artificial intelligence technology.”

Read it at The Hill → https://thehill.com/homenews/senate/6092827-gop-thom-tillis-donald-trum…

MSNBC — independent

Progressives launch new push to reject campaign money from AI industry

The money: progressive candidates and groups pushing back on a bipartisan pro-AI super PAC backed by top tech leaders.

Nearly a dozen Democratic congressional candidates are banding together to call on campaigns to refuse money from a major pro-artificial intelligence super PAC ahead of the November midterms. The new pledge targets Leading the Future, a bipartisan organization that has raised at least $140 million since it launched last year. Its financial supporters include the likes of OpenAI co-founder Greg Brockman, Trump-aligned Palantir co-founder Joe Lonsdale and the creators of venture capital firm Andreessen Horowitz. The initiative, first reported by MS NOW, comes at a time of heightened scrutiny of

Read it at MSNBC → https://www.ms.now/news/progressives-campaign-money-ai-industry-anthrop…

The Hill — independent

Sanders: ‘Embarrassing’ how ‘ignorant’ Trump is on AI

Sanders calls the president "ignorant" about the dangers of AI — the sharpest attack we hold, and the headline is built from his adjectives.

Sen. Bernie Sanders (I-Vt.) went after President Trump on the issue of artificial intelligence (AI), saying he has a lack of knowledge around its risks. “Look, it is embarrassing, really, I think, going beyond politics, how ignorant this president is about the dangers of AI,” Sanders told MS NOW’s Akayla Gardner in a clip posted…

Headline built from one side's words: the critic's adjectives carry the headline unquoted-in-body form — “Sanders: ‘Embarrassing’ how ‘ignorant’ Trump is on AI”

Read it at The Hill → https://thehill.com/policy/technology/6092611-sanders-criticizes-trump-…

The Hill — independent

63 percent believe AI could destroy humanity: Survey

The only poll figure anyone publishes — 63 percent — with no sample, dates or question wording in the text.

A majority of Americans are concerned that artificial intelligence poses grave risks to the future of humanity, according to a new poll released on Wednesday. The POLITICO survey found that 63 percent of U.S. adults believe this technology could someday destroy mankind. Twenty-six percent of respondents said there was only a “moderate risk” of this…

Method behind a number: a poll figure is published with no sample, date or method in the text — “The POLITICO survey found that 63 percent of U.S. adults believe this technology could someday destroy mankind.”

Poll reported without method: a single percentage carries the headline; no sample, no dates, no question wording — “The POLITICO survey found that 63 percent of U.S. adults believe this technology could someday destroy mankind.”

Read it at The Hill → https://thehill.com/policy/technology/6092503-poll-ai-humanity-risks/

The Atlantic — independent

Destroying Humanity Is Against the Law

Puts it as liability rather than risk: hold developers accountable before a mass-extinction event, not after.

Ideas Destroying Humanity Is Against the Law Don’t wait for a mass-extinction event before holding AI developers accountable. By Juliette Kayyem Illustration by The Atlantic. Sources: Brendan Smialowski / AFP / Getty; Julien De Rosa /AFP / Getty; Matt Ramey / AFP / Getty. September 16, 2026, 9:38 AM ET Share Save This weekend, the AI executives Sam Altman, Elon Musk, and Dario Amodei agreed with the need to slow the pace of AI development. In light of the potentially apocalyptic capabilities of the products they have created, the AI titans would like a little more time to figure out how to saf

Read it at The Atlantic → https://www.theatlantic.com/ideas/2026/09/ai-risk-criminal-law/688615/?…

The Hill — independent

AI is finally making us question our bias in favor of deregulation

Dates the turn: Amodei's essay on Saturday, his CBS remark on Sunday, and the Lieu–Moran proposal it lines up with.

On Saturday, Anthropic CEO Dario Amodei published an essay urging tech leaders to slow the pace of AI development. On a Sunday CBS News interview, he added that it “could be a good idea” to have an AI kill switch — not unlike what Rep. Ted Lieu (D-Calif.) and Rep. Nathaniel Moran (R-Texas) have proposed. “Some…

Read it at The Hill → https://thehill.com/opinion/technology/6090742-anthropic-ceo-ai-killswi…

National Review — independent

Against AI Doomerism

Argues the case against caution in the outlet's own collective voice, with nobody named as holding the view.

Whose voice: first-person editorial voice, no actor and no source — “We can’t afford to fall behind.”

First person plural, no actor: the report argues in a collective voice with nobody named as holding the view — “We can’t afford to fall behind.”

Read it at National Review → https://www.nationalreview.com/2026/09/against-ai-doomerism/

The New York Times — independent

The Subtle Sign That A.I. Is Not Yet a National Political Issue

The counterweight to the 63 percent: a Times/Siena poll finds voters still uncertain, including on which party would handle AI better.

A Times/Siena poll shows many voters are still uncertain on the issue, including which party would do a better job on it.

Read it at The New York Times → https://www.nytimes.com/2026/09/16/upshot/poll-ai-midterms-voters.html

The New York Times — independent

Mark Zuckerberg Takes Aim at Anthropic in Debate Over A.I. Slowdown

The only report of a rival lab pushing back: Meta's chief executive appearing to jab at Anthropic over the slowdown call.

On social media, Meta’s chief executive appeared to jab at Anthropic by saying that leading A.I. labs should be focused on safety rather than improving their own technology.

Inference hedged twice: a motive is read off a post and marked as reading, not reporting — “On social media, Meta’s chief executive appeared to jab at Anthropic by saying that leading A.I. labs should be focused on safety rather than improving their own technology.”

Read it at The New York Times → https://www.nytimes.com/2026/09/15/technology/mark-zuckerberg-anthropic…

The Atlantic — independent

The Only Thing Capable of Uniting Bernie and Bannon

The alignment that cuts across the parties — Sanders and Bannon both fearing AI could wipe out humanity.

Politics The Only Thing Capable of Uniting Bernie and Bannon They both fear that AI could wipe out humanity. By Toluse Olorunnipa Finn Gomez / Getty September 15, 2026, 8:09 PM ET Share Save If there was any question over whether the concept of “the enemy of my enemy is my friend” is still operative in Washington’s hyper-partisan political environment, the answer may be found in the budding coalition of progressives and MAGA-friendly activists who have united in recent days to speak out against artificial intelligence. Which is how Senator Bernie Sanders and Steve Bannon found themselves speak

Read it at The Atlantic → https://www.theatlantic.com/politics/2026/09/steve-bannon-bernie-sander…

Checked, didn’t cover

Mother Jones

Reuters N NPR P Politico T The Washington Post

Breitbart

Silence is recorded, not inferred — each of these was read when we built this story and carried nothing on this event.

We could not read these

The Wall Street Journal

Associated Press

These are must-cite outlets for this desk that we did not manage to read when we built this story. That is a gap in our reach, not a silence on their part, and it is not evidence of anything about this story.

Read another desk

Sections

Registers

Desks

Trump calls the AI extinction fear a "hoax"; a Republican senator answers "Come on, it's not a hoax"

What is being argued about

What each side gets called

Two polls, no method shown

Claims and verdicts

Who wrote it

Coverage, outlet by outlet