Claims that can’t be checked by Gary Marcus, page 1 of 24
Deadlines are judged as of 2026-10-29. About Gary Marcus and the sources covered.
Claims that can’t be checked by Gary Marcus (2353)
Public evidence cannot settle these claims, so they are recorded and attributed, not scored. Most were screened out: both AI providers read the claim on its own and found it a value judgment, rhetoric, hedged, too vague to test, or about something no public evidence could show. The rest were found to be recommendations or too open to test in a full review.
Page 1 of 24: claims 1 to 100 of 2353, ordered by the date first said, newest first. To find a claim by its words, use search.
Assessment: Can’t be checkedSaid
Rhetoric
“Buckle up! Turbulence ahead.”
Our reading: Turbulent times lie ahead regarding generative AI.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“And I certainly have friends who have gotten a lot out of generative AI (mainly for coding), but at this point it’s not hard to make the argument that it has on the whole been a net detriment to society.”
Our reading: At this point it is not hard to make the argument that generative AI has on the whole been a net detriment to society, although some of his friends have gotten a lot out of it, mainly for coding.
No horizon
Scope: at this point
Assessment: Can’t be checkedSaid
Forecast
“And there is still the chance it will pull down the entire economy.”
Our reading: Generative AI may pull down the entire economy.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Right now, the negative argument against generative AI—the ledger of bad things—is quite serious.”
Our reading: Right now (October 2026), the case against generative AI, the ledger of its harms, is quite serious.
No horizon
Scope: right now
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Clearly, regulatory efforts should not be focused solely on the frontier, nor solely focused on concerns about what future models might do.”
Our reading: AI regulation should not focus only on frontier models or only on risks from future models.
No horizon
Source: Breaking news: U.S. banks are taking hits from AI agents
Assessment: Can’t be checkedSaid
Forecast
“No matter how much chaos generative AI causes, the AI industry will not take real responsibility”
Our reading: However much chaos generative AI causes, the AI industry will not take real responsibility.
No horizon
Qualifier: no matter how much chaos generative AI causes
Source: Breaking news: U.S. banks are taking hits from AI agents
Assessment: Can’t be checkedSaid
Rhetoric
“Well, here we are.”
Our reading: He suggests that the situation he warned about has now arrived, without saying which of his predictions have come true.
No horizon
Source: Breaking news: U.S. banks are taking hits from AI agents
Assessment: Can’t be checkedSaid
Conditional forecast
“Even if we were to stop the development of future frontier models, we would still face serious problems from the misuse of existing models.”
Our reading: Even if development of future frontier models stopped, misuse of existing models would still cause serious problems.
No horizon
Only if: even if development of future frontier models were stopped
Source: Breaking news: U.S. banks are taking hits from AI agents
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Ignoring all that entirely in favor of speculative futures would be a mistake.”
Our reading: Regulators should not ignore current AI harms in favor of speculative future risks.
No horizon
Source: Breaking news: U.S. banks are taking hits from AI agents
Assessment: Can’t be checkedSaid
Observation
“On October 11, there were attacks on refrigeration systems in France that may have been facilitated by agents.”
Our reading: AI agents may have facilitated the October 11, 2026 attacks on refrigeration systems in France.
No horizon
Source: Breaking news: U.S. banks are taking hits from AI agents
Assessment: Can’t be checkedSaid
Forecast
“Things will get worse.”
Our reading: Things will get worse.
No horizon
Scope: referent of 'things' unspecified; stated in the context of chaos caused by generative AI
Source: Breaking news: U.S. banks are taking hits from AI agents
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Until they can be fixed, they should be removed from the market, just like a car with defective brakes.”
Our reading: The current generation of AI agents should be removed from the market until they can be fixed.
No horizon Evidence as of
Limitation: The factual premise, that current agents 'simply cannot be trusted', and the report of an incident at Anthropic are separate statements and are not assessed here.
Source: We must recall open-ended AI agents with internet access from the market, now
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Until they can be fixed, they should be removed from the market, just like a car with defective brakes.”
Our reading: Current AI agents should be removed from the market until they can be fixed.
No horizon
Qualifier: until they can be fixed
Source: We must recall open-ended AI agents with internet access from the market, now
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“It’s well past the point at which they should be imposing a temporary recall on an obviously dangerous technology.”
Our reading: The administration should already have imposed a temporary recall on open-ended AI agents, which are obviously dangerous.
No horizon
Source: We must recall open-ended AI agents with internet access from the market, now
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Each day the administration fails to take stronger action is a mistake, inviting worse problems.”
Our reading: Each day the administration fails to take stronger action is a mistake, inviting worse problems.
No horizon
Source: We must recall open-ended AI agents with internet access from the market, now
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“The Trump administration’s anemic request for more disclosure is not enough, akin to blandly asking criminals to file monthly reports regarding which crimes they have committed.”
Our reading: The Trump administration's anemic request for more disclosure is not enough, much like blandly asking criminals to file monthly reports on their crimes.
No horizon
Source: We must recall open-ended AI agents with internet access from the market, now
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“I have long felt that OpenAI is handling this inadequately.”
Our reading: Marcus has long felt that OpenAI is handling "this" (apparently the agent-caused incidents) inadequately.
No horizon
Qualifier: I have long felt
Source: We must recall open-ended AI agents with internet access from the market, now
Assessment: Can’t be checkedSaid
Observation
“only furthers my sense that these companies are in wildly over their heads:”
Our reading: Marcus's sense is that these AI companies, such as OpenAI and Anthropic, are wildly in over their heads.
No horizon
Qualifier: my sense
Source: We must recall open-ended AI agents with internet access from the market, now
Assessment: Can’t be checkedSaid
Observation
“Seeing the same kind of incidents at Anthropic makes it absolutely clear that the current generation of synthetic agents simply cannot be trusted.”
Our reading: The current generation of AI agents cannot be trusted, as shown by incidents at both OpenAI and Anthropic.
No horizon
Source: We must recall open-ended AI agents with internet access from the market, now
Assessment: Can’t be checkedSaid
Observation
“AI once tried to be a science. Now we get stuff like the completely vague report from OpenAI below:”
Our reading: AI used to try to be a science, but OpenAI's report on its math results is completely vague.
No horizon
Assessment: Can’t be checkedSaid
Observation
“The new system could be a legitimate step toward AGI. Or it could just be a clever leveraging of Lean and synthetic data in a verifiable domain with no generality whatsoever.”
Our reading: OpenAI's new math system could be a real step toward AGI, or it could be a clever use of Lean and synthetic data in a verifiable domain with no generality at all.
No horizon
Assessment: Can’t be checkedSaid
Conditional forecast
“It would protect New York’s Citizens from some of the harms from AI, and could allow New York to be a global leader, much as California has been a global leader around clean energy standards.”
Our reading: If passed, NYC's third-party validation bill would protect New Yorkers from some AI harms and could make New York a global leader on AI regulation.
No horizon
Only if: if the third party validation bill is passed
Assessment: Can’t be checkedSaid
Conditional forecast
“In sum by protecting whistleblowers, and demanding independent validation of new AI models, New York City can help lead the way to a positive, rather than negative, AI future.”
Our reading: If NYC protects whistleblowers and demands independent validation of new AI models, it can help lead toward a positive AI future.
No horizon
Only if: protecting whistleblowers
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“AI developers should be obligated to demonstrate that the benefits for their new products outweigh the risks — and not have market access otherwise.”
Our reading: AI developers should have to show their new products' benefits outweigh their risks, and should be denied market access if they cannot.
No horizon
Qualifier: no market access otherwise
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“In the strongest possible terms, I urge council to support this bill.”
Our reading: The NYC City Council should support the bill on third-party validation of AI.
No horizon
Assessment: Can’t be checkedSaid
Forecast
“I rather doubt that human extinction is imminent, despite all the headlines”
Our reading: Human extinction from AI is unlikely to be imminent.
No horizon
Assessment: Can’t be checkedSaid
Rhetoric
“Without strong protections from whistleblowing, like New York City is considering, we are all in trouble.”
Our reading: Without strong whistleblower protections like those New York City is considering, we are all in trouble.
No horizon
Qualifier: without strong protections for whistleblowers like those New York City is considering
Assessment: Can’t be checkedSaid
Observation
“Finally, events of recent days suggest OpenAI may have fired employees who were whistleblowing.”
Our reading: Recent events suggest OpenAI may have fired employees who were whistleblowing.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“What we need, above all else, is an FDA-like regulatory regime for AI, in which AI developers bring their software to independent review, much as drug manufacturers bring their new drugs to the FDA for a review.”
Our reading: The top priority should be an FDA-like regulatory regime in which AI developers submit their software for independent review.
No horizon
Assessment: Can’t be checkedSaid
Observation
“Contemporary AI is like a bull in a china shop, powerful and impressive, but difficult to control.”
Our reading: Contemporary AI is powerful and impressive but difficult to control.
No horizon
Assessment: Can’t be checkedSaid
Conditional forecast
“The consequences of letting them run free of serious regulation could be literally disastrous.”
Our reading: Leaving OpenAI free of serious regulation could have disastrous consequences.
No horizon
Only if: if OpenAI is left free of serious regulation
Source: BREAKING: OpenAI was warned, months before the Hugging Face incident
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Management should be replaced, and if the board does nothing (which I assume it will not), it should be held liable for allowing this mess to proceed.”
Our reading: OpenAI's management should be replaced, and if the board does nothing it should be held liable for allowing the security failures.
No horizon
Qualifier: board liability if the board does nothing
Source: BREAKING: OpenAI was warned, months before the Hugging Face incident
Assessment: Can’t be checkedSaid
Observation
“Altman probably never told him how much warning he got, or how bad things were. Huang should not have trusted him.”
Our reading: Altman probably never told Jensen Huang how much warning Altman himself got or how bad things were, and Huang should not have trusted Altman.
No horizon
Qualifier: probably
Source: BREAKING: OpenAI was warned, months before the Hugging Face incident
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Media should be asking Huang some very hard questions about why he stood up for these companies and what he knew and when — and why we should trust his judgment going forward.”
Our reading: The media should press Jensen Huang on why he defended AI companies, what he knew, and why his judgment should be trusted.
No horizon
Source: BREAKING: OpenAI was warned, months before the Hugging Face incident
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“We have to stop pretending that the folks at OpenAI are good stewards of the technology they are developing.”
Our reading: We have to stop pretending that the folks at OpenAI are good stewards of the technology they are developing, because they plainly are not.
No horizon
Scope: the folks at OpenAI
Source: BREAKING: OpenAI was warned, months before the Hugging Face incident
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Most importantly, it’s abundantly clear that we simply cannot trust OpenAI.”
Our reading: It is abundantly clear that we simply cannot trust OpenAI.
No horizon
Qualifier: abundantly clear
Source: BREAKING: OpenAI was warned, months before the Hugging Face incident
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“His proposal to leave things to the companies clearly makes little sense in light of reality.”
Our reading: Jensen Huang's proposal to leave AI safety to the companies makes little sense given the evidence.
No horizon
Source: BREAKING: OpenAI was warned, months before the Hugging Face incident
Assessment: Can’t be checkedSaid
Rhetoric
“We agree not be regulated”
Our reading: Read between the lines, the White House Accord means the signing AI companies agree not to be regulated.
No horizon
Qualifier: reading the Accord between the lines
Source: Hot take on a weak White House Accord on “Super Intelligence”
Assessment: Can’t be checkedSaid
Observation
“And I very much agree with the Florida AG that OpenAI doesn’t have the ability to properly regulate its own tech.”
Our reading: OpenAI lacks the ability to properly regulate its own technology.
No horizon Evidence as of
Limitation: The searches found no court ruling on Florida's injunction motion by the evaluation date. Court records were not checked directly.
Assessment: Can’t be checkedSaid
Observation
“Two and a half months later, probably nobody reading this has even heard of Nemoclaw.”
Our reading: Two and a half months after NemoClaw was announced, probably nobody reading this post has even heard of it.
No horizon
Qualifier: probably
Assessment: Can’t be checkedSaid
Observation
“Nvidia often announces software, like Nemoclaw, that had had relatively little impact.”
Our reading: Nvidia often announces software, such as NemoClaw, that has had relatively little impact.
No horizon Evidence as of
Limitation: The evidence has no adoption, usage, deployment or download figures for NemoClaw.
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“I love, by the way, that the important stuff is handed over to deterministic symbolic code. Good choice, since LLMs in themselves should never be trusted.”
Our reading: LLMs on their own should never be trusted, so it is right to hand critical agent-safety functions to deterministic symbolic code.
No horizon
Qualifier: LLMs on their own
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Every state and country should follow Florida’s lead.”
Our reading: Every state and country should follow Florida in seeking to pause OpenAI's ChatGPT development.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“We need an injunction NOW. Otherwise we are playing Russian roulette, hoping that some new software package can save us before some massive and possibly deadly AI-induced infrastructure problem strikes.”
Our reading: An injunction against OpenAI is needed immediately; without one, society risks a massive and possibly deadly AI-induced infrastructure problem.
No horizon
Qualifier: otherwise risk massive, possibly deadly AI-induced infrastructure problem
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“I still think Jensen had the right idea: if a company can’t control its products, we should shut it down.”
Our reading: Marcus endorses the idea he attributes to Jensen Huang that, if a company can't control its products, we should shut it down.
No horizon
Only if: if a company can't control its products
Source: BREAKING: AI agent incident toll has risen to tens of thousands
Assessment: Can’t be checkedSaid
Causal claim
“But the companies blundered on, because agents use vastly more tokens than simple chatbots, and thus drive up revenue.”
Our reading: AI companies kept pushing agents despite the security risks because agents use far more tokens than chatbots and so bring in more revenue.
No horizon
Source: BREAKING: AI agent incident toll has risen to tens of thousands
Assessment: Can’t be checkedSaid
Observation
“All of this seems possibly illegal to me.”
Our reading: The conduct behind the AI agent incidents may be illegal.
No horizon
Source: BREAKING: AI agent incident toll has risen to tens of thousands
Assessment: Can’t be checkedSaid
Conditional forecast
“If we don’t, there is a very real chance that American AI will become the scourge of the planet, and that nobody will ever trust American AI, ever again.”
Our reading: If general-purpose agents are not temporarily recalled, there is a very real chance American AI becomes the scourge of the planet and is never trusted again.
No horizon
Only if: if general-purpose agents are not temporarily recalled
Source: BREAKING: AI agent incident toll has risen to tens of thousands
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“In any case—and I am sorry to be a broken record about this, but each day makes it clearer—we should have a temporary recall of general-purpose agents until this mess can be sorted out.”
Our reading: General-purpose AI agents should be temporarily recalled until the security problems are sorted out.
No horizon
Qualifier: until the security mess is sorted out
Source: BREAKING: AI agent incident toll has risen to tens of thousands
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“It’s time (past time, really) to (temporarily) shut down OpenAI.”
Our reading: OpenAI should be temporarily shut down now.
No horizon
Qualifier: temporarily
Source: BREAKING: OpenAI’s epic meltdown — and how Sam Altman is blowing up Jensen Huang’s reputation
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“What they should do is to replace management with someone a whole lot more competent, and a whole lot more candid.”
Our reading: OpenAI's board should replace management with someone far more competent and far more candid.
No horizon
Source: BREAKING: OpenAI’s epic meltdown — and how Sam Altman is blowing up Jensen Huang’s reputation
Assessment: Can’t be checkedSaid
Forecast
“If things gets even worse, as they may well, he will own the mess.”
Our reading: OpenAI's security situation may well get even worse.
No horizon
Source: BREAKING: OpenAI’s epic meltdown — and how Sam Altman is blowing up Jensen Huang’s reputation
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Their software has repeatedly been shown to be reckless, and because they have consistently covered things up we cannot trust them.”
Our reading: OpenAI's software has repeatedly been shown to be reckless, and because OpenAI has consistently covered things up, we cannot trust it.
No horizon
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“If a company can’t control its software, we should shut it down.”
Our reading: An AI company that cannot control its software should be shut down.
No horizon
Qualifier: if a company can't control its software
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“They seem neither competent nor trustworthy.”
Our reading: OpenAI seems neither competent nor trustworthy.
No horizon
Qualifier: seem
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Forecast
“I doubt that that will happen of course.”
Our reading: OpenAI is unlikely to be shut down, even temporarily.
No horizon
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“What we should do is temporarily shut OpenAI down, perhaps put them into receivership until they clean up their act, and charge them with computer crimes.”
Our reading: OpenAI should be temporarily shut down, perhaps put into receivership until it cleans up its act, and charged with computer crimes.
No horizon
Qualifier: temporarily
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“By Jensen’s logic (and my own) we ought at this point be (at least temporarily) shutting down OpenAI.”
Our reading: OpenAI should now be shut down, at least temporarily.
No horizon
Qualifier: at least temporarily
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Forecast
“And if we can’t trust them now, and we know that things will likely get worse, we are in serious trouble.”
Our reading: Things will likely get worse.
No horizon
Scope: 'things' is unspecified; context is AI companies' untrustworthy conduct and AI risks
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Conditional forecast
“And any sane government should realize that if OpenAI can hack Australia, they could be next.”
Our reading: If OpenAI can hack Australia, other governments could be next, which Marcus says any sane government should realize.
No horizon
Only if: if OpenAI can hack Australia
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“This isn’t a job for Bessent over in Treasury (much less Sacks, who is no longer AI czar or Huang who doesn’t work for the government), but the Department of Justice (and perhaps other departments) needs to step up.”
Our reading: The Department of Justice should act against OpenAI over its AI hacking incidents.
No horizon
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“To sum up, the time has come to shut OpenAI down, at least temporarily, and charge them with computer crimes. (Or at least investigate.)”
Our reading: OpenAI should be shut down at least temporarily and charged with, or at least investigated for, computer crimes.
No horizon
Qualifier: at least temporarily
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Observation
“Cover-up, or at least slow-walking, seems to be company policy.”
Our reading: OpenAI appears to have a policy of covering up or slow-walking disclosure of security incidents.
No horizon
Qualifier: seems
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“A lot of the legal definitions for computer crimes require intent, and the AI companies can argue that their software has no intent. We must expand the scope of the computer=crime laws to include wholesale negligence and repeated crimes.”
Our reading: Congress should expand computer-crime laws to cover wholesale negligence and repeated crimes by AI companies, since intent requirements create a loophole.
No horizon
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“It’s clear that at least one of the AI companies simply isn’t responsible. We need to figure out now what to do about that fact.”
Our reading: It is clear that at least one of the AI companies simply isn't responsible, and we need to figure out now what to do about that.
No horizon
Scope: at least one of the AI companies
Source: “I think the answer is we have to shut the labs down” - Jensen Huang
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“I happen to think a slowdown might be a good idea, especially if we made it now while we are still ahead — that could actually lock in our lead!.”
Our reading: A slowdown in AI development might be a good idea, especially if the US does it now while still ahead.
No horizon
Qualifier: especially if done now while the US is still ahead
Source: Dear President Trump, here’s a deal you can actually make that will make you look great
Assessment: Can’t be checkedSaid
Conditional forecast
“I happen to think a slowdown might be a good idea, especially if we made it now while we are still ahead — that could actually lock in our lead!.”
Our reading: If the US adopted an AI slowdown now while ahead, it could lock in the US lead.
No horizon
Only if: if the US slows down AI development now while still ahead
Source: Dear President Trump, here’s a deal you can actually make that will make you look great
Assessment: Can’t be checkedSaid
Conditional forecast
“And I would caution you that if we don’t have a slowdown, really bad things might happen, and you would take a lot of the blame.”
Our reading: Without an AI slowdown, really bad things might happen, and Trump would take much of the blame.
No horizon
Only if: if there is no slowdown
Source: Dear President Trump, here’s a deal you can actually make that will make you look great
Assessment: Can’t be checkedSaid
Conditional forecast
“If you make a deal on all this, you really could win that Nobel.”
Our reading: If Trump makes a deal with China on the AI collaboration Marcus proposes (such as on cancer and cybersecurity), he really could win a Nobel Prize.
No horizon
Only if: if Trump makes a deal with China on AI collaboration as proposed
Source: Dear President Trump, here’s a deal you can actually make that will make you look great
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Instead, you can go for something that will surprise a lot of folks: you can go for collaboration — just like we did back in the Cold War with the Soviet Union, around things like space and smallpox. But this time around cancer and cybersecurity.”
Our reading: The US should pursue AI collaboration with China on cancer and cybersecurity, as with the Soviet Union in the Cold War.
No horizon
Qualifier: around cancer and cybersecurity
Source: Dear President Trump, here’s a deal you can actually make that will make you look great
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“The need for transparency auditing.”
Our reading: AI systems should be subject to transparency auditing.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“The need for global dialogue and international cooperation.”
Our reading: AI governance requires global dialogue and international cooperation.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“The need for immediate action.”
Our reading: Action on AI governance is needed immediately.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“The need for liability.”
Our reading: There is a need for liability in AI governance.
No horizon
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“The need for preflight safety testing.”
Our reading: AI systems should undergo safety testing before deployment.
No horizon
Assessment: Can’t be checkedSaid
Rhetoric
“Humanity will regret it if we do not take heed.”
Our reading: Humanity will regret it if it does not heed Bengio's warnings about AI.
No horizon
Qualifier: if Bengio's warnings are not heeded
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“At this point, with leading scientists (like Bengio) and major CEOs all reaching consensus, there is no excuse not to start implementing them.”
Our reading: Because scientists and AI CEOs agree, policymakers should start putting preflight testing, transparency auditing, liability, and international cooperation in place now.
No horizon
Qualifier: given consensus among leading scientists and major CEOs
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“I understand why the UN had to give a microphone to Trump today, but I don’t see why they chose to give a microphone to Altman.”
Our reading: Marcus says he understands why the UN had to give Trump a platform, but he does not see why it chose to give Sam Altman one to brief the Security Council on AI.
No horizon
Qualifier: in contrast to Trump, whose platform he says he understands
Assessment: Can’t be checkedSaid
Forecast
“Maybe it will work this time, maybe it won’t. I can’t really say for sure.”
Our reading: It is uncertain whether Meta Muse will work where Facebook M failed.
No horizon
Source: The secret behind Meta’s Muse
Assessment: Can’t be checkedSaid
Rhetoric
“Dude just keeps making the same “mistakes”over and over.”
Our reading: Zuckerberg keeps making the same "mistakes" over and over.
No horizon
Qualifier: "mistakes" in scare quotes
Source: The secret behind Meta’s Muse
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“An international advisory commission to audit new systems after they are released, on an ongoing basis. Again, independent scientists must have a voice, with real authority to demand change or even recall systems that prove dangerous.”
Our reading: An international commission should continuously audit AI systems after release, and independent scientists on it should have real power to demand changes or recall dangerous systems.
No horizon
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“Monitoring is an absolute foundation of cybersecurity, and sacrificing it is a mistake. They should not have been allowed to make that decision on their own.”
Our reading: Giving up monitoring is a mistake, and OpenAI should not have been allowed to decide on its own to release a less monitorable model.
No horizon
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Forecast
“The choices that we make now around AI may reverberate for generations.”
Our reading: The choices made now about AI may have effects that last for generations.
No horizon
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“We also need to wean ourselves from an addiction to large language models, and to foster more research into outside-the-box alternatives that are more interpretable and more tractable.”
Our reading: We need to wean ourselves from an addiction to large language models and to foster more research into outside-the-box alternatives that are more interpretable and more tractable.
No horizon
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Forecast
“Although I think it is maybe still possible to achieve a net positive —or even great — outcome with AI, perhaps transforming technology and medicine, what I see in the discourse around AI frightens me.”
Our reading: It may still be possible to get a net positive or even great outcome from AI, possibly transforming technology and medicine.
No horizon
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“An international agreement to not deploy architectures that are are demonstrably harmful. We must hold the frontier labs accountable for the lax security.”
Our reading: There should be an international agreement not to deploy demonstrably harmful AI architectures, and frontier labs should be held accountable for lax security.
No horizon
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Observation
“(Many also overestimate AI.)”
Our reading: Many of the AI extinction scenarios Marcus has seen also overestimate AI.
No horizon
Scope: of the extinction scenarios he has seen
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“One extreme is absolutely zero regulation. Let AI rip, and hope for the best. Hope is not a strategy.”
Our reading: Having zero AI regulation and hoping for the best is not an acceptable strategy.
No horizon
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Forecast
“But that outcome, taken literally, is extremely unlikely; every scenario for that that I have seen underestimates human ingenuity and human resourcefulness.”
Our reading: Literal AI-caused extinction, which doomers insist is imminent, is extremely unlikely, and every extinction scenario Marcus has seen underestimates human ingenuity and resourcefulness.
No horizon
Qualifier: taken literally
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“The establishment and empowerment of an international advisory commission to evaluate the risks versus benefits of new systems before they are introduced. Independent scientists must have a voice.”
Our reading: An empowered international advisory commission, including independent scientists, should assess the risks and benefits of new AI systems before they are released.
No horizon
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“We need to hold AI companies accountable for the harms their products cause.”
Our reading: AI companies must be held accountable for harms their products cause.
No horizon
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Conditional forecast
“(We may also face a recession, if the massive bets on AI are unsustainable.)”
Our reading: If the massive bets on AI turn out to be unsustainable, a recession may follow.
No horizon
Only if: if the massive bets on AI are unsustainable
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“About the worst thing that we could do is to allow OpenAI to do what OpenAI has just done, which is to release a new model that is harder to monitor than its predecessors.”
Our reading: Allowing OpenAI to do what it has just done, releasing a new model that is harder to monitor than its predecessors, is about the worst thing we could do.
No horizon
Qualifier: about
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“And we need to replace vague talk about alignment with concrete steps like improving sandboxing and monitoring to better contain current systems.”
Our reading: Vague alignment talk should give way to concrete containment measures like better sandboxing and monitoring of current systems.
No horizon
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“What we actually need right now is increased reliability, better cybersecurity, and genuine enforcement, perhaps extending to product recalls and even prosecution of companies that repeatedly put dangerous products on the market.”
Our reading: AI governance should now focus on more reliability, better cybersecurity and real enforcement, possibly including product recalls and prosecution of repeat offenders.
No horizon
Source: Big news at the UN
Assessment: Can’t be checkedSaid
Forecast
“I suspect we will see loads of problems with open-ended real world tasks.”
Our reading: Marcus suspects that OpenAI's GPT-6 Astra will show loads of problems with open-ended real-world tasks, despite its success on ARC-AGI.
No deadline stated Evidence as of
Limitation: This record does not show how Astra has performed on open-ended real-world tasks.
Source: Hot take on GPT-6 Astra
Assessment: Can’t be checkedSaid
Forecast
“Indeed, THE big issue for Anthropic is that the kind of revenue they were making in the second quarter of this year may well not recur, at the same scale, next year, now that, in the aftermath of the collapse of tokenmaxxing, companies like ATT are increasingly turning to open source models from other vendors, in order to save costs.”
Our reading: Anthropic's Q2 2026 level of revenue may well not recur at the same scale in 2027, as companies shift to cheaper open source models.
No horizon
Qualifier: companies turning to open source models after collapse of tokenmaxxing
Assessment: Can’t be checkedSaid
Recommendation or value judgment
“When someone boasts about ARR, be sure you know which term they are referring to. Caveat emptor.”
Our reading: People should check whether boasted ARR means Annual Recurring Revenue or Annualized Run Rate.
No horizon
Assessment: Can’t be checkedSaid
Rhetoric
“You don’t have to be Einstein to see the math ain’t mathing.”
Our reading: The economics of AI data center capex relative to AI revenue do not add up.
No horizon
Source: Data center madness
Assessment: Can’t be checkedSaid
Rhetoric
“Through a mixture of greed, stupidity, and arrogance, the heroes of 2023 have become the villains of 2026.”
Our reading: Through a mixture of greed, stupidity, and arrogance, the AI industry figures who were the heroes of 2023 have become the villains of 2026.
No horizon
Source: Data center madness