Pressure of Truth
Exposing the spin on all sides of the news.
Finance

Chip and AI Stocks Fall in Seoul, Tokyo and U.S. Premarket After Amodei, Altman and Musk Call for Slowing Frontier AI Development

Anthropic CEO Dario Amodei published an essay on Sept. 12, 2026 urging labs to "pace the frontier"; Sam Altman and Elon Musk backed it, and AI-linked shares fell on Monday, Sept. 14.

How spun is the coverage?Coverage bias 4.3 / 10
5 sides analyzed16 sources cited

Three Rivals Agree, and the Market Falls Anyway

On Sept. 12, 2026, Anthropic CEO Dario Amodei posted an essay on X with an unusual request. He asked the handful of companies building the world's most advanced AI models to deliberately slow down how fast they make those models smarter[1][2].

Two of his biggest rivals said he was right. Elon Musk posted "Dario is right." Sam Altman, who runs OpenAI, wrote that he agreed the industry needs to "pace the frontier"[1][7]. These are executives who have sued and publicly feuded with each other before.

Two days later, on Monday, Sept. 14, AI-linked stocks fell hard. SK Hynix closed down more than 6% in Seoul, and Samsung Electronics dropped more than 4%[5][6]. SoftBank, a major OpenAI investor, fell 10% in Tokyo[6]. Nasdaq futures were off 1.8%, with Nvidia down about 2.9% before the U.S. market even opened[4][6].

Here is the tension a reader has to sit with. The people calling for caution are the same people who stand to gain the most if their rivals slow down too. And the market reaction wasn't as simple as it looked, either — those Asian chip stocks had just hit record highs days earlier[13][14]. Both things are true at once, and untangling them is the whole story.

What Amodei Is Actually Asking For

Amodei's essay lays out three specific steps, not a vague call for caution[1][3]. First, he wants outside safety testers to get "employee-level access" to AI models. Right now, most outside evaluators test a finished product through a limited interface, similar to how a regular customer would use it. Amodei wants them inside the building instead, able to see a model before it's released and run the same deep tests engineers run[1][3].

Second, he wants the leading labs to agree on shared safety standards among themselves. Third, he wants those labs to start coordinating with governments[1][3].

The reason he gives is specific and time-stamped. Amodei warns that within 6 to 12 months, a swarm of autonomous AI agents working together could effectively hijack large parts of the internet, functioning like a botnet, and cause "potentially hundreds of billions of dollars in damage"[1]. That same day, Altman separately said 2026 would be "an ill-advised moment" for OpenAI to go public, pushing a long-anticipated IPO to at least 2027[1][4].

Notably, Amodei's essay says the plan is designed so it does not cost the U.S. its "commercial advantage" or its lead in the AI race[1]. He is arguing safety and speed don't have to trade off against each other.

Who Actually Gains From Everyone Slowing Down

Strip away the specifics and a structural fact sits underneath this whole debate: whoever is already ahead benefits most from a pause[9]. Anthropic and OpenAI are, by most measures, at the front of the pack. A voluntary agreement to slow down freezes the current order — it locks in their lead while everyone else, including startups and open-source developers trying to catch up, loses ground too.

There's a second structural point that cuts the same way. A safety program built for a company that already has a legal team and a dedicated evaluations department is a manageable cost for that company. For a smaller startup with no compliance department, the very same rules can be a serious burden[9]. That gap between who can easily absorb a rule and who can't is the mechanical core of an argument you'll hear a lot in this story: regulatory capture, meaning rules written by an industry that end up protecting that industry from competition.

None of this means Amodei's safety concern isn't sincere. It just means the same proposal can be both a genuine warning and a move that happens to help the people making it. Those two things aren't mutually exclusive, and untangling them from the outside is hard.

The Adviser Calling It a Cartel

That's the argument David Sacks makes, and he makes it with some authority. Sacks stepped down in March 2026 from his formal role as the White House's AI and crypto czar, after hitting a 130-day limit for that kind of position, but he still co-chairs the President's Council of Advisors on Science and Technology and remains closely aligned with the administration's AI policy[9].

Sacks calls Amodei's plan "regulatory capture." He argues that if Anthropic and OpenAI genuinely believe a model is too dangerous to release, they can simply choose not to release it — no new framework required. Turning something they can already do on their own into a request for outside rules, in his view, is what makes it suspicious. He put it bluntly, saying the labs were acting as if "antitrust law has to be suspended so you can form a cartel"[9].

Sacks also went after the credibility of METR, the outside evaluator Anthropic relies on for safety testing, arguing it's too closely tied to Anthropic's investors and staff to act as a neutral referee[9].

His position matters for another reason: since he's an influential Trump-aligned voice on tech policy, his public skepticism signals that the kind of government coordination Amodei is asking for probably isn't coming from this administration anytime soon[9].

Beijing Hears a Different Argument Entirely

China's government isn't arguing about safety testing at all. Its objection is to a different part of the essay — the parts aimed at restricting China's access to AI technology[10][11].

The state-run Global Times called Amodei's essay a "Cold War playbook," arguing its real goal is to box in China's AI progress through tighter chip export controls and what it called "regulatory monopolies," while keeping China out of any say in how AI gets governed globally[10][11]. China's Foreign Ministry also pushed back publicly[10].

One flashpoint is a technique called distillation, where a smaller AI model is trained to copy the outputs of a larger, more expensive one. It lets a lab approach cutting-edge performance without paying the full cost of building it from scratch — which is part of how Chinese labs have kept pace despite existing chip restrictions. Cracking down on distillation, alongside tighter export controls, is exactly the kind of coordination Amodei's third step envisions, and it's exactly what Beijing objects to[10][11].

Even Amodei has acknowledged the problem with his own framework: he's called China the "toughest dilemma" for the plan, since any pact that only Western labs join doesn't actually slow down the global frontier — it just changes who builds the next model[8].

A Chip Investor's View, and What's Still Unsettled

For chip makers like Samsung, SK Hynix, Nvidia and Micron, this isn't an argument to weigh in on — it's a risk to absorb. Their business depends heavily on demand for high-bandwidth memory, specialized chips that sit next to an AI processor and feed it data fast enough to keep up. That demand is set by how many, and how large, the next generation of frontier models turns out to be. A credible signal that model growth will slow down hits those stocks immediately, even before any actual sales change[4][5][12].

That's part of why the numbers moved the way they did on Sept. 14. But the drop needs context most headlines left out: Korean chip shares had just set records days before, with the KOSPI hitting an all-time high and Samsung's market value crossing $1 trillion[13][14]. Coverage varied in how it handled that. CNN's report on the slide didn't mention the recent record highs[5][13]. Forbes led with a specific number — SoftBank chief Masayoshi Son's fortune dropping by $8 billion — a real figure that also makes a single day's premarket dip feel more dramatic than it may turn out to be[6]. The Washington Post's headline described the essay as a call for "oversight," which reads as more sympathetic than the essay's actual text, given that it describes a voluntary industry pact, not a regulatory one[2].

What actually happened, stripped down, is narrow: one essay, two public endorsements, and one trading session where AI-linked stocks fell[1][4][5][6]. No lab has announced a halt to its work. No government has proposed a rule. Whether Monday's drop reflects a real shift in how investors see the AI buildout, or just one volatile session following a record run, isn't something a single day can answer.

Like this article?

Share this article

The Bias Ledger average rating 4.3

The same story, as framed by outlets across the spectrum, ordered least to most biased. The bias score (1 = straight, 10 = heavily spun) is an AI assessment of that framing — click an outlet to see its track record. The tell is the word choice or omission that reveals the angle.

OutletVantageBiasHow they frame itThe tell
CNBCU.S. center, business2"OpenAI rules out IPO this year as Altman, Musk & Amodei warn AI is moving too fast" and "AI stocks slide after Anthropic, OpenAI CEOs urge slowdown"[1][4].Leads with the market and deal angle. The IPO delay is bundled into the safety headline, which quietly implies a causal link the reporting does not establish. Otherwise close to straight reporting, and it separately ran the China dilemma angle[8].
CNNU.S. center-left3"AI stocks slide after top industry CEOs call for slowdown of technology's development"[5]."After" does the causal work without asserting causation. The piece stacks the day's percentage drops without noting that Korean chip shares were falling from record highs set days earlier — an omission that makes the slide look larger than the trend[13][14].
The Washington PostU.S. left-of-center4"Anthropic's Amodei calls for AI oversight, joined by Altman and Musk"[2].The word "oversight" reframes a voluntary industry pact as a call for public accountability — a more sympathetic reading than the essay's text supports. The rare-alliance framing foregrounds credibility; the conflict-of-interest objection gets less room.
Fox NewsU.S. right4"Anthropic CEO Dario Amodei calls to slow AI arms race as critics push back"[7]."Arms race" and "as critics push back" are both in the headline, so the rebuttal arrives before the proposal does. The framing centers the incumbents-versus-startups fight rather than the risk claim itself.
ForbesU.S. center-right, business5"AI Stocks Slide In Premarket Amid Push For Slowdown In Development" — with a subhead noting SoftBank chief Masayoshi Son's fortune dropped by $8 billion[6].Attaches a single billionaire's paper loss to a market story. It is a real number, but it personalizes a sector move and makes a one-day premarket dip feel like a rout.
Global TimesChinese state media, published under the People's Daily8Frames the essay as "packed with containment provisions targeting China" and "in essence a 'Cold War playbook' for the AI sector"[10][11].Answers a safety argument entirely on geopolitical grounds and never engages the botnet claim. The quoted experts are unnamed and uniformly hostile — "inappropriate," "groundless" — which is the standard signature of a coordinated state response.

References

  1. OpenAI rules out IPO this year as Altman, Musk & Amodei warn AI is moving too fast — CNBC · U.S. business news, owned by Comcast/NBCUniversal; market-centric framing
  2. Anthropic's Amodei calls for AI oversight, joined by Altman and Musk — The Washington Post · U.S. left-of-center; owned by Jeff Bezos, who has separate AI and cloud interests
  3. Anthropic, OpenAI CEOs call for slowdown in AI development — Axios · U.S. center; insider-access model, brevity-formatted
  4. AI stocks slide after Anthropic, OpenAI CEOs urge slowdown — CNBC · U.S. business news, Comcast/NBCUniversal
  5. AI stocks slide after top industry CEOs call for slowdown of technology's development — CNN · U.S. center-left; owned by Warner Bros. Discovery
  6. AI Stocks Slide In Premarket Amid Push For Slowdown In Development — Forbes · U.S. center-right business media; majority-owned by Integrated Whale Media
  7. Anthropic CEO Dario Amodei calls to slow AI arms race as critics push back — Fox News · U.S. right; owned by Fox Corporation
  8. Anthropic's Amodei says China presents 'toughest dilemma' for his proposed AI slowdown — CNBC · U.S. business news, Comcast/NBCUniversal
  9. David Sacks blasts OpenAI and Anthropic over slowdown — Neowin · Independent technology trade site; aggregates and quotes primary posts
  10. US-China cold war over AI? China slams Anthropic's call to slow AI development — The Week · Indian news magazine (Malayala Manorama group); relaying Chinese state media and MOFA statements
  11. China's Global Times Says US's AI Concerns Are A Ploy To Slow China's AI Progress — OfficeChai · Indian tech trade blog; summarizing Global Times, which is Chinese state media under People's Daily
  12. Chip stocks shed more than $1 trillion as selloff hits companies powering AI boom — CNBC · U.S. business news, Comcast/NBCUniversal
  13. S.Korea's KOSPI hits record high as SK Hynix, Samsung extend AI-fueled rally — Investing.com · Commercial financial data and news portal; trader-oriented
  14. South Korea Kospi bull market: SK Hynix, Samsung, AI trade — CNBC · U.S. business news, Comcast/NBCUniversal
  15. OpenAI boss Sam Altman spells out how and why the AI industry wants to slow down: 'We could lose control' — CNBC · U.S. business news, Comcast/NBCUniversal
  16. Palantir stock may win even if AI boom slows: why Wall Street sees 50% upside — Invezz · Commercial investing news site; promotional/retail-trader orientation