Gebru and Bender: AI companies' summer hype collapses under expert scrutiny
Researchers argue breakthroughs in vulnerability detection, math, and superintelligence claims are marketing, not science.
What to know
- Major 2026 AI claims—from vulnerability detection to mathematical breakthroughs to superintelligence warnings—do not withstand scrutiny by domain experts, according to researchers Gebru and Bender.
- Mathematicians have accused OpenAI of research misconduct and plagiarism in its Astra breakthrough claims; cybersecurity experts attribute hacking incidents to corporate negligence, not rogue AI.
- The researchers argue AI companies use anthropomorphizing language to market products as superhuman while evading accountability, and that these narratives are ideological rather than scientific.
“Claims of incipient, dangerous superintelligence are not based in good scientific or engineering practice. Rather, they are narratives based in ideologies of transhumanism, eugenics, and wishful thinking about imagined future digital humans.”
Timnit Gebru and Emily Bender, AI researchers · MIT Technology Review ↗ · Sep 21
Timnit Gebru AI researcher and co-authorEmily Bender AI researcher and co-authorOpenAI AI company making claimsAnthropic AI company making claimsJacob Coxon Anthropic engineerTristan Buckmaster Math professor, NYU Courant Institute
How it unfolded 1 development · click the chart to see its coverage articlesposts
-
1
Gebru and Bender publish critique: claims don't withstand expert examination
Timnit Gebru and Emily Bender published an analysis arguing that all major AI industry claims from the summer—vulnerability detection, mathematical breakthroughs, and superintelligence warnings—collapse under scrutiny by domain experts. They argue the companies use anthropomorphizing language to evade accountability and that hacking incidents reflect OpenAI negligence, not rogue AI.
“Breathless claims about AGI and new capabilities fall apart pretty quickly under scrutiny.”
— Timnit Gebru -
first by Mastodon, 1d ago · also MIT Tech Review, MIT Technology Review
-
E
From @ timnitGebru and me on what we should take from the 2026 summer of AI hype: https://www. technologyreview.com/2026/09/2 2/1144867/dont-be-fooled-summer-ai-hype/
2 more of the top 3 · 27 posts in this stretch
-
It's a tough situation when your claim to fame is "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?" arguing models should be kept to < GPT-2 size because larger models will serve no purpose or function:Text generated by an LM is not grounded in communicative intent, any model of the world, or any model of the reader’s state…
-
G
It's so good to see this in the MIT Tech Review. Let's hope the message gets through. 🤞Big props to @ timnitGebru & @ emilymbender for putting such a fine point on things! "Instead of OpenAI being prosecuted for creating malware that hacked another company, press releases, news outlets, media personalities, and lawmakers refer to “rogue models”…
-
-
background
NYU math professor Buckmaster accuses OpenAI of research misconduct and plagiarism — Tristan Buckmaster, a math professor at NYU's Courant Institute, published a statement suggesting OpenAI had stolen other people's work and improperly attributed it, two days before OpenAI's mathematical breakthrough claim.
-
background
Anthropic engineer Jacob Coxon departs, warns of superintelligence race — Anthropic engineer Jacob Coxon went viral announcing his departure from the company, claiming Anthropic and OpenAI are racing toward self-improving superintelligence.
-
background
Anthropic and OpenAI claim mathematical breakthroughs — Anthropic announced a model had made a mathematical breakthrough; OpenAI soon claimed a mathematical breakthrough of its own using its Astra chatbot.
-
background
OpenAI–Hugging Face hacking incident disclosed; Anthropic and Meta follow with their own — An OpenAI–Hugging Face hacking incident occurred, followed by Anthropic (proudly) and Meta (reluctantly) disclosing similar incidents involving their models.
-
background
Anthropic claims Claude Mythos finds vulnerabilities better than security experts — Anthropic released a claim that its Claude Mythos model outperforms most security experts at finding software vulnerabilities.
Also covered reported alongside — the timeline has no entry for these yet
-
first by Mastodon, 1d ago · also HN Frontpage, MIT Tech Review
1 more headline
- The Download: why AI's latest breakthroughs and fears may be more hype than rea HN Frontpage · 19h ago
What people are saying 21 voices from 2 sites · best of 34 · verbatim
- Yesterday
-
"Seriously, I don't get what sort of world these people are living in."Perhap it's a world that does not believe in "demons"Otherwise known by those seeking to escape it as "the real world"
-
Amusing juxtaposition:>Here’s why they say we must remain skeptical about AI hype.…22 nations have called for a new global body to oversee AI
-
T
By me and @ emilymbender . "According to the AI industry, we should be more worried about a fictional machine god than about the climate catastrophe these data centers exacerbate, the asthma suffered by those living near them, the rising electricity bills of the public subsidizing them, or the water...[for] cooling them." https://www…
-
A must read about «AI» hype by @ emilymbender and @ timnitGebru . https://www. technologyreview.com/2026/09/2 2/1144867/dont-be-fooled-summer-ai-hype/ I found it particularly instructive to read the Leiden Declaration by mathematicians on «AI»: https:// leidendeclaration.ai/ I'd like to see a similar declaration on # hamradio and # AI . There's…
-
And if they'd discovered how to summon unicorns from Candy Gumdrop Mountain, we'd all be "OMG UNICORNS!"But no one has summoned a demon.No one has built an actual, independent, thinking-for-itself, sci-fi AI.They've built some very interesting tools that can be used for some very interesting, and sometimes useful, purposes. They then started…
-
Your analogy assumes that no one ever has summoned demons before and that the demons would be predisposed to rob banks.A closer analogy to what's happening would be someone trains a monkey to steal jewellery and then is shocked when the monkey steals jewellery from their neighbors when they told it to steal from their own shop.Clearly all…
-
It's just a marketing stunt. I can't get Claude to align buttons properly most of the times, they're not going to conquer the world
-
I think the pumped-up hype nicely coincides with some A.I. companies' desire to go public in the coming weeks or months.I see lots of fabricated news on the internet concerning LLMs. Like one that claims GPT-6 broke an Enigma enciphered message which has withstood decrypting for almost 80 years. And how this seasoned cryptographer stood in awe…
-
People aren't seeing the forest through the trees.While lots of developers, journalists, analysts, investors and influencers are bickering about AGI, goalposts, benchmarks, hypes and fears, entire markets are being transformed silently and steadily.I don't program anymore. (Massive change)I removed tons of technical debt (Massive change, lol)I am…
-
Feels like the blockchain bubble all over again, just with more impressive demo reels. Still waiting for my LLM to write perfect code.
-
That's not what the LLM hacking accidents have been like at all though. To improve the analogy, it would be like summoning a totally passive demon, giving it weapons and placing it next to a bank, place a small fence around it, and then command it to perform a totally safe "exercise" that is exactly like robbing a real bank.LLMs do not have…
-
If I run a port scanner / vulnerability fuzzer pointed at your network that churns for days and eventually cracks something and breaks your system, would you be upset with me, or my bash loops?What if you tried to sue me for damages and my defense was, "Your honor, this was a highly advanced AI gone rogue, I couldn't possibly be held negligent, no…
-
Is this meant to link to the article as opposed to a newsletter that mentions the article?Article link is:
- Sep 22
-
R
I also think we should be waiting for the experts' opinion - just not exclusively the opinions of those who have an existential fear of their expertise being made obsolete... Between big tech hype and criticism for its own sake, we should be able to adopt an intellectually honest position regarding the abilities and limitations of LLMs…
-
B
Not for one second to let #AI & #DataCentres off the hook but there is a far greater, vastly more certain threat. #ClimateChange @fluidsguru.bsky.social@bsky.brid.gy
-
In the MIT Technology Review , @ timnitGebru @ emilymbender puncture recent LLM hype and remind us to wait for experts to take a sober second look
-
D
> Describing them as “superintelligence” or “rogue models” ascribes agency to products rather than to the companies building them. This framing markets these companies’ products as “superhuman” and, at the same time, helps the companies evade accountability for their actions. > Instead of OpenAI being prosecuted for creating malware that hacked…
- Sep 21
-
P
TY @sundaedivine.lol@bsky.brid.gy for nailing it! According to Google's #AI , women don't count: bsky.app/profile/paul... W/ #ALText : #StrongerTogether due to #inclusion :) #PresidentThiel , #PresidentVought , #PresidentEllison , #PresidentMiller et al. are #Christiban . They'll destroy #democracy , if not stopped. RE:
- Sep 20
-
P
NO MENTION OF WATER CONSUMPTION RE: #AI ? No mention of desalination plants? No mention of ice caps melting quicker from AI? Only #energy #Auspol #climate #rorts RE: https://bsky.app/profile/did:plc:ebyirfmcml2wmydrexhxu27t/post/3mvyomanskk26 Cartoon Beaver Eek
-
I
#AI #datacenters #computerchips #chips #carbonfootprint #energy #power #fossilfuels #co2 #emissions #climatedisaster #SimonClark explains #NVIDIA to the #BackRows youtube.com/shorts/TllgS... #climate #climatechange #ClimateEmergency #ClimateCrisis #globalWarming #ExtremeWeather #polycrisis Speedrun Nvidia's climate talk...
- Sep 18
-
Hey #AI crowd. I do #biodiversity work; here's what an evolutionary biologist sees in multi-agent AI. RE: