KPMG issued a report citing all the transformational ways GenAI has transformed industry, it’s been widely cited.
One minor problem: it turns they used AI to write the report, and it made up all of the evidence.
KPMG have now withdrawn the report in full.
https://www.ft.com/content/b3828e92-4961-4b39-84f0-c42f33be3c3f
Sounds like an entire engagement team is heading "on the beach".
@That_Damn_Frank @GossiTheDog unlikely, sadly, because it has done its job. More likely they will get their sales bonus. Lots of business development, lots of new 'AI' clients because 'UBS uses AI', zero publicity or admission outside of bubbles like ours about how the whole thing was a fraud.
Similar to the UK's Advertising Standards Authority (ASA) telling a company to 'not show an ad again' that is clearly misleading. There is no obligation to issue a retraction or to apply anything like the same publicity to the fact it was a lie, just a tacit admission that _that exact ad_ won't run again. (A similar, equally misleading ad will of course be hot on its heels.)
@GossiTheDog because of *course* they did.
@GossiTheDog is his where we head with the reliabiloty of the worlds third line of defence? This is dystopian. Nothing matters, financial statements get partly checked only, stock reports might include hallucinations.
@compliance @GossiTheDog No one reads those reports anyway so if AI generates them with even more crap in them, no one will notice. But the reports will be absorbed in the database for more AI to digest.
What could go wrong?
@bouriquet @compliance @GossiTheDog Don’t forget: AI is generating the snackable summaries as well 😎
@bouriquet @compliance @GossiTheDog
100% what they wanted. Says they pulled the report from "some" of their websites. So they appease some skeptics and fools. Rinse repeat, and soon the lies are homogenized.
Similar to AI spitting out art-slop for prints 24/7/365 and putting it in front of people to find out what is ignored or accepted. Then deleting the art-slop least shared or sold.
@GossiTheDog More people need to hear this! I can't believe the amount of BS in the news about "AI".
@GossiTheDog This has gone so far that I think we're going to have to change the old "It was DNS" to "It was AI hallucinations"
@GossiTheDog Time between publication and the issues getting flagged… 😅
@Kensan A very old problem. Over a century ago, Mark Twain noted that a lie gets around the whole world before the truth can get its shoes on, and he wasn't by any means the first person to note that. I'd bet good money that some Ancient Greek wag said the same.
@GossiTheDog I just wanted to sign up to read it "1€ for 4 weeks", sounds fine
"Then €69 per month." WOAH, okay, important to cancel early enough!!
Jun 12, 2026, 11:06 AM@fgraf @GossiTheDog FYI you can usually read FT articles by searching for the URL in archive.is eg https://archive.is/X4zGK
@GossiTheDog 😂 99% of big four can be replaced by AI.. with proper guardrails
RE: https://cyberplace.social/@GossiTheDog/116736495067438407
@GossiTheDog
beyond parody
I have a hunch about SpaceX's SEC filing… 😑
@GossiTheDog collective hallucination aka bullshit
@GossiTheDog
We really need to get rid of the term "Hallucinations". It's an algorithm spewing nonsense. It's a bullshit generator spewing bullshit. It's a chatbot trained to serve up crap that sounds plausible. In short, it's a tool that does exactly what it's designed to do.
And the people who published this monstrosity should not be able to hide behind the tool they used!
@GossiTheDog
@jarjan +1
Terminology related to Actual Idiocy has to change drastically.
@OrangeR And it does, constantly. In fact, this particular sector of human language is probably the most aggressively dynamic of any. If you're a legal adult, you already know a lot of words for 'stupid', and that's why.
@GossiTheDog @jarjan
As expressed before , no term related to thinking or feeling should be associated with the algorithmic vomit (©️ Seb ) these systems spew .
@jarjan I doubt that actually knowledgeable people misunderstand it, and you do have to use SOME term to describe the phenomenon. Nearly all modern language is allegorical, so unless you just invent completely new words, all terms are going to be metaphors. And in cases like this, any term you come up with is going to be misunderstood and misused by people who AREN'T knowledgeable anyway, so what difference does it make?
@wesdym
The point is that the term "hallucinating" is grossly misleading (possibly intentionally so) for most people. It assigns a human / intelligence quality to a statistical language generator.
The correct term would be: errors
It's really not complicated, but "errors" doesn't sound so nice.
@jarjan I'm willing to bet that you have not polled "most people" and have approximately zero evidence to back up this claim that I'm sure you believe very much.
You're going to have to eventually learn to live with the world not always going your way, especially when 1) you can't do much or anything about it, and 2) your frustration stems from things you imagine rather than what you can be reasonably sure is true based on good evidence. Your life will be better.
Let that process start today.
@jarjan @GossiTheDog that horse seems well bolted unfortunately.
Jun 12, 2026, 10:48 AM@GossiTheDog [Nelson Muntz laughter]
@GossiTheDog 🤪😂
So some will continue to trust this kind of company ?
@OrangeR Yes. Big = 'good' or 'strong' or 'reliable' or 'trustworthy' is a primitive but powerful human heuristic, baked into our evolved neurology. (Which is optimized for survival around a quarter million years ago, not in context of the often contrary artifices of what we quaintly call 'civilization'.) This is why companies work so hard to project that sense. I'm not trying to be cynical, clever, or cute here. This is just a fact of human neurology, and thus of human behaviour.
@GossiTheDog AI doesn't hallucinate. That is just a marketing ploy to try and suggest it is in some way conscious (it isn't and never will be). It is the people unthinkingly using it that are deluded.
@sothach I admit I'm fuzzy on the exact meaning of the term in this context, but I don't believe any acknowledgeable people believe that AI is conscious, and the term is used and intended metaphorically, not literally. Indeed, it stems from the fact that AI does not and CAN not reason: It formulates responces based on statistical patterns in its training data, which if poorly curated (as is very often the case) may be fictional or non-sensical.
I love this for them.
@GossiTheDog Kind of makes one wonder whether the consulting firms have a future if they are just relying on AI to produce their own reports. OTOH, this story isn't exactly a ringing endorsement of that replacement.
Jun 12, 2026, 11:06 AM@briankrebs@infosec.exchange @GossiTheDog@cyberplace.social You would have thought that the large, er, "thought leading" consulting firms would have been violently anti-AI; verbose sycophancy as a service is basically the whole business model. That and pre-arranged supplier approval with large orgs so that nobody manages to hire any consultants who are actual experts.
@mavnn @briankrebs @GossiTheDog "sycophancy as a service" brilliant take, I'm going to borrow that.
Jun 12, 2026, 11:25 AM@shelldozer@oldbytes.space Thank you, although I have to honest and say that I think I've heard it used before. Unfortunately I can't remember where, so I'll take a bit of Schrödinger pride in the case that maybe I did coin it on the fly...
@shelldozer @mavnn @briankrebs @GossiTheDog
OK. What's a good business name that somehow involves "Derp" and/or "Herp" that offers boutique sycophancy services that don't use AI?
Herp de Derp Enterprises doesn't AI.
What's a good rate? Should it charge by the hour or by ego milestone?
You think there would be a good market pairing with "Derpdy Dew Lawn Services"? Maybe a package deal?
@briankrebs @GossiTheDog consulting firms are hosed
"The industry" needs to realize that using LLMs for things that have a binary works/doesn't work scenario (development, some exploit work) is vastly different from producing text that cannot be verified the same way.
AI is the wet dream of most of the big consulting firms. why pay expensive SMEs when you can use AI and still charge the end customer obscene amounts? way higher profit margin.
every time i've ever had to deal with the big consulting firms, i walk away feeling like i need to bathe in bleach.
@paul_ipv6 @briankrebs @GossiTheDog
If these consulting firms want to continue charging high rates, wouldn't it be better to remain silent about using AI tools? It will damage their reputation and earnings.
@0815 @briankrebs @GossiTheDog
sadly, no. these firms will continue to get fed and pentagon money regardless of probity or performance...
@GossiTheDog Excellent. Brilliant even.
LOL
@GossiTheDog The only fittingly shambolic sequel is where UBS learns that they have a shadow IT problem and, empirically, they have integrated AI agents across investment advisory, risk management and compliance monitoring; and maybe having someone's pet openclaw setup handling risk and compliance is not good.
@GossiTheDog par for the course
@GossiTheDog 'AI is great!' says report written by an AI for an audience of people who really want AI to replace employees. It's impressive how they've found a way to be dumber than the 'billionaires are great' op eds, written by millionaires paid by billionaires, published in papers owned by billionaires.
Jun 12, 2026, 11:49 AM@GossiTheDog And that's why we pay big bucks to these big consulting companies, because they are really big and they are really good at consulting and research.
Oh...again?
@GossiTheDog
Their "AI" appears to have been even more successful at making up stuff than their humans used to be (don't ask for my evidence on the reliability of KPMG reports, I signed some kind of NDA).
They always did good presentations and excellent lunches, though, according to the senior managers of their clients.
@GossiTheDog ...the wrong conclusion is that AI will not fundamentally change things, the right conclusion is that KPMG and EY are totally unreliable companies that shouldn't be hired after this and earlier scandals.
@GossiTheDog "KPMG issued a report citing all the transformational ways GenAI has transformed industry, it’s been widely cited.
One minor problem: it turns they used AI to write the report, and it made up all of the evidence. "
You could not make this stuff up. At least, I couldn't. Just not enough imagination.
So Great! Made me laugh out loud.
@lemgandi @GossiTheDog Me too!
@GossiTheDog The AI fox hired to write a report on why the AI fox excels at guarding the actual henhouse. Funded entirely by actual hungry greedy trillionaire foxes.
This is so remarkably stupid, and utterly predictable. Musk is betting on "too big to fail" to keep him out of jail and afloat. Let's promise to Bernie Madoff and ENRON Musk and all his ilk.
@GossiTheDog 😂
Ancrer toute affirmation dans source identifiable.
Distinguer le validé, le probable, le spéculatif et l’infondé.
Croiser les disciplines et les époques.
Interroger les biais des institutions et des consensus.
Maintenir une humilité épistémique ne jamais confondre l’absence de preuve avec la preuve de l’absence.
Projeter réflexion dans le futur, intégrant l’incertitude.
méthode
Socrate, Descartes, Montaigne, Kant doit être réactivée à chaque génération les Py se réincarnent
@GossiTheDog Another reason why A.I. is a FRAUD.
@GossiTheDog Ironically, if it stops people relying on biased KPMG analysis that would be fairly transformative
