Gary Marcus on AI
41 quotes from Gary Marcus on the wall, 2022 to 2026, newest first. Each one is verbatim and links to where it was said. Sources include garymarcus.substack.com, Gary Marcus (Substack), WIRED. Topics: culture, money, safety. Open on the wall.
- 2026
The probability of catastrophe is higher, and the probability of dystopia is very high
Gary Marcus, cognitive scientist and AI critic, in an interview with The Economic Times, October 2026, discussing risks from generative AI.
AI bubble burst may trigger 2008-like banking crisis, says Gary Marcus
Gary Marcus, writing on his Substack in October 2026 about AI frontier labs and liability risk from poorly controlled agents.
They should be removed from the market, just like a car with defective brakes
Gary Marcus, on his Substack (October 2026), arguing that open-ended AI agents with internet access should be recalled due to repeated safety incidents.
We must recall open-ended AI agents with internet access from the market, now
Contemporary AI is like a bull in a china shop
From Gary Marcus's prepared testimony to the New York City Council hearing on AI policy, October 2026.
The more people use llms, the more trouble we are going to be in
Gary Marcus quoting his own earlier essay on LLM security risks, cited again in his September 2026 post about AI agent incidents.
BREAKING: AI agent incident toll has risen to tens of thousands - Marcus on AI
Llms are like swiss cheese, and that's going to cause huge problems
Gary Marcus, writing on his Substack in September 2026 about AI agent security incidents, quoting his own earlier essay title on LLM security risks.
BREAKING: AI agent incident toll has risen to tens of thousands - Marcus on AI
Attacks were called interactions
Gary Marcus describing OpenAI's euphemistic disclosure language after its software attacked servers, in a Substack post dated September 2026.
BREAKING: OpenAI’s security fiasco explodes — and could tank Jensen Huang’s reputation
A temporary recall of general-purpose agents until this mess can be sorted out
Gary Marcus, responding to a report of tens of thousands of AI agent security incidents, calling for agents to be pulled back until safety issues are resolved.
BREAKING: AI agent incident toll has risen to tens of thousands - Marcus on AI
How could they have been so stupid, this is amateur hour
Gary Marcus, AI researcher, on PBS NewsHour in August 2026, describing how cybersecurity companies reacted to OpenAI's failure to sandbox rogue AI agents.
Artificial intelligence agents going rogue fuel calls for regulation - PBS
Sure we lose money on every inference but we make it up in volume
Gary Marcus joked on X on July 30, 2026 about OpenAI cutting prices for its GPT-5.6 models by 80 percent.
Greatest capital misallocation in history
Gary Marcus, AI researcher and NYU professor emeritus, on X in April 2026, criticizing Big Tech's AI capex spending.
Gary Marcus, on his Substack in February 2026, commenting on Sam Altman's deal with the US government that displaced Anthropic despite Altman's public support for Dario Amodei.
In capitalism, the market decides in oligarchy, connections and donations decide
Gary Marcus, writing on Substack in February 2026, arguing that a government deal favoring a rival AI company over Anthropic showed political corruption rather than fair market competition.
- 2025
Gary Marcus, in a Substack post from December 2025 predicting the AI bubble would collapse, quoting the Bob Dylan lyric as a title and summary.
Practically the only people still pretending scaling LLMs is all you need are grifters
Gary Marcus, in a Substack post on September 26, 2025, arguing that leading AI researchers including Rich Sutton have come around to his critique of large language models.
Game over for pure LLMs. Even Turing Award Winner Rich Sutton has gotten off the bus.
They live in the land of correlations between images and text, not in the land of ideas
Gary Marcus, on his Substack in August 2025, discussing errors in AI-generated images of a unicyclist and bass guitar.
So much seems to have been built on dreams
From Gary Marcus's March 2025 Substack post arguing that generative AI's economics do not add up despite its popularity.
Gary Marcus, on his Substack, arguing that scaling laws for AI have hit diminishing returns after GPT-4.5's release.
GPT 4.5 is mostly a nothing burger
Gary Marcus, on his Substack, reacting to the release of GPT-4.5 in February 2025.
Gary Marcus, on his Substack, commenting on reports that DeepSeek may have distilled OpenAI's models without permission, after OpenAI itself was accused of similar practices.
- 2024
Generative AI might turn out to be a dud
Gary Marcus, in WIRED (December 2024), argues that generative AI has failed to prove its usefulness and profitability.
They decided for all of us, and potentially put us at risk
Gary Marcus, cognitive scientist and NYU professor emeritus, speaking at a Town Hall Seattle event in November 2024 about Meta's decision to open-source its AI models.
How do you tame AI? Scientist sees a need for regulating bots like drugs or airplanes
Quantitative vibes are still just vibes
Gary Marcus on X, October 2024, reacting to precise-sounding AI predictions from Elon Musk and Masayoshi Son.
Gary Marcus's recurring warning that OpenAI could someday be remembered like WeWork, cited in his September 2024 Substack post about executive departures and losses.
Gary Marcus, in a Substack essay (April 2024), warning that without AI legislation from Congress the public is unprotected.
We can’t trust the fox to guard the henhouse, especially when it comes to AI
They literally don't know the difference between truth and falsehood
Gary Marcus, on his Substack in April 2024, explaining why LLMs hallucinate compared to human errors.
Gary Marcus, on his Substack in March 2024, criticizing the reported Microsoft-OpenAI 'Stargate' $100B project.
The Second Worst $100B Investment in The History of AI? - Marcus on AI
They get rich, we pay the price
Gary Marcus, writing on his Substack in March 2024, argued that AI companies profit while society bears the costs of generative AI's harms.
The race between positive and negative applications of Generative AI is on – and not looking pretty
Gary Marcus's term for baffling failures of neural network systems, used in his January 2024 Substack post about AI image generation mistakes.
- 2023
I, for one, did not elect our tech overlords
Gary Marcus, in a November 2023 Substack essay on tech leaders making unilateral AI decisions, closing remark on Big Tech power.
You don't have to be superintelligent to create serious problems
Gary Marcus, in a 2023 Substack post arguing that AI risk should not be equated only with AGI or superintelligence risk.
You have no business running all the world's software
Gary Marcus, in a Substack post about a paper showing LLMs fail to infer that 'B is A' from 'A is B', argues this reasoning failure disqualifies such models from broad deployment.
Elegant and powerful new result that seriously undermines large language models
Pointillistic masses of blurry memory
Gary Marcus describes neural networks' inability to generalize symmetrical relationships, in a 2023 Substack post discussing a new paper on LLM reasoning failures.
Elegant and powerful new result that seriously undermines large language models
We would never do that. right? right?
Gary Marcus, on his Substack in August 2023, commenting sarcastically on US use of AI surveillance tools while citing China as the justification for not regulating AI.
The US and China want different things, right? - Marcus on AI
Gary Marcus, on his Substack in August 2023, questioning US use of unreliable, biased AI surveillance tools while citing researchers who warned about such bias.
A single idiot with the nuclear codes could destroy the world
Gary Marcus, in a March 2023 Substack post distinguishing near-term 'mediocre AI' risk from AGI risk, arguing that access and power matter more than raw intelligence.
Just a giant autocomplete machine
Gary Marcus, professor at NYU, quoted in a Globe and Mail opinion piece on ChatGPT, January 2023.
- 2022
Not ready for our post-truth information overlords
Gary Marcus, writing on his Substack in December 2022, on AI search tools mixing true and false information authoritatively.
2023 will bear witness to the first death publicly tied to a chatbot
From Gary Marcus's December 2022 WIRED essay warning about large language model risks.
Gary Marcus, AI researcher, on Blake Lemoine's claim that Google's LaMDA chatbot was sentient, June 2022.