KPMG Pressed Its Auditor To Pass on AI Cost Savings (ft.com) 33
An anonymous reader shares a report: KPMG, one of the world's largest auditors of public and private companies, negotiated lower fees from its own accountant by arguing that AI will make it cheaper to do the work, according to people familiar with the matter. The Big Four firm told its auditor, Grant Thornton UK, it should pass on cost savings from the rollout of AI and threatened to find a new accountant if it did not agree to a significant fee reduction, the people said.
The discussions last year came amid an industry-wide debate about the impact of new technology on audit firms' business and traditional pricing models. Firms have invested heavily in AI to speed up the planning of audits and automate routine tasks, but it is not yet clear if this will generate savings that are passed on to clients.
Grant Thornton is auditor to KPMG International, the UK-based umbrella organisation that co-ordinates the work of KPMG's independent, locally owned partnerships around the world. Talks with Grant Thornton were led by Michaela Peisger, a longtime audit partner and executive from KPMG's German member firm, who became KPMG International's chief financial officer at the beginning of 2025.
The discussions last year came amid an industry-wide debate about the impact of new technology on audit firms' business and traditional pricing models. Firms have invested heavily in AI to speed up the planning of audits and automate routine tasks, but it is not yet clear if this will generate savings that are passed on to clients.
Grant Thornton is auditor to KPMG International, the UK-based umbrella organisation that co-ordinates the work of KPMG's independent, locally owned partnerships around the world. Talks with Grant Thornton were led by Michaela Peisger, a longtime audit partner and executive from KPMG's German member firm, who became KPMG International's chief financial officer at the beginning of 2025.
Of course (Score:2)
How else is your consulting division going to sell A.I. to clients if your own audit figures don't back them up?
Goosing yourself, for the Gander. (Score:5, Insightful)
..it should pass on cost savings from the rollout of AI and threatened to find a new accountant if it did not agree to a significant fee reduction..
I estimate KPMG customers pulling this same excuse against KPMG, by the end of this sentence.
Come to think of it, why do we even need human auditors at KPMG again?
Yes, KPMG. Be careful what you fuck others over for. You just might earn it.
Re: (Score:3)
Come to think of it, why do we even need KPMG again?
FTFY just a bit.
Re: (Score:2)
Just waiting for it to turn out that AI cooked their books or missed some massive fraud, and they try to claim it's not their fault.
Re: (Score:2)
Come to think of it, why do we even need human auditors at KPMG again?
KPMG auditors are human? I guess I missed that news story!
Of all of the things to use AI for... (Score:1)
Re: Of all of the things to use AI for... (Score:1)
So, like humans then?
Re: (Score:3)
Re: (Score:2)
No.
This "it's okay that AI makes mistakes because humans aren't perfect either" is ridiculous. The kinds of mistakes humans make are so far removed from the kinds of 'mistakes' LLMs make that it's misleading to even call them mistakes.
Humans, for example, will never accidentally summarize a document that don't exist.
LLMs are not science fiction robots. They are not 'electronic brains' that think and reason. The very idea is absurd on its face.
Re: (Score:2)
Indeed. While humans also hallucinate occasionally and many are prone to really deranged cult-like behavior (see the "pedo-protectors" in the US, for example), humans will very rarely make the type of mistakes that LLMs make. Because LLM hallucinations look good and well researched while being total fabrications. Humans may deliver low-quality results, but they will be visibly low quality. Humans may also deliver high-quality fakes, but they will only do that intentionally (and hence often criminally, with
Re: (Score:2)
Re: (Score:2)
I also wonder who professionals will feel signing off on AI work since they would then be legally responsible for it.
Senior people will refuse and may outright leave. Junior people may get coerced into it. But guess what, there are seniority requirements for audit jobs, exactly for that reason. And once caught, an auditor can look for a job in a different field. Nobody will hire them again as auditor.
\o/ (Score:1)
They sound like douchebags - was the intention to encourage the rejection?
Auditing must be AI-free (Score:4, Interesting)
Auditing is about establishing the truth about the situation within the company. Which is why AI should never be allowed to be used in the context of auditing. Or at the very least the auditors should never rely on the AI-generated results provided by the company. What's stopping companies from telling AI to generate a set of data that would make them pass an audit, hiding the real facts? Auditors should rely on facts, not confabulations, and should dig as deep as necessary to uncover all the dirt that's hiding under the surface.
Re: (Score:2)
Afterthought:
Which is why auditors should actually be asking for more money because now they've got to cut through even more corporate BS.
Re: (Score:2)
Re: (Score:2)
The auditor is the one using AI. One of the things AI is good at is looking at lots of boring data and finding weird patterns.
Auditors used to have to go through and check every transaction to see if someone screwed up or was fudging the numbers. Modern accounting practices plus software can pretty much do that automatically now, leaving the auditors to look for more subtle stuff. AI can help with that more subtle stuff.
Re: (Score:2)
One of the things AI is good at is looking at lots of boring data and finding weird patterns.
I hear people blindly repeat that particular 'fact' pretty often, but it does not mean what people seem to think it means. I've also seen the claim stated as "neural networks are good at spotting patterns in data".
When people hear things like that, they imagine some hyper-intelligent science fiction robot spotting patterns in seemingly random data. They might even think that modern LLMs can be used to analyze large amounts of data and find hidden patterns. This is silly nonsense.
When we say that 'AI is g
Re: (Score:2)
I'm not sure where you're going with any of that. You might want to google "unsupervised learning" though.
Re: (Score:2)
I'm not sure where you're going with any of that.
That's because you're stupid.
You might want to google "unsupervised learning" though.
You should take your own advice. You have no idea just how stupid your comment looks here.
Re: (Score:2)
Lol. A compelling argument. At least nobody will be mistaking you for an AI.
Re: (Score:2)
What argument? You said something irrelevant. I agreed that you don't understand my post and gave you some much needed advice.
If it doesn't hurt to be you, it should.
Re: (Score:2)
So you could say there's a whole industry to fill in for gutless spineless managers.
Don't shoot me, I'm just an independent messenger.
Oh goody another way AI will crash our economy (Score:3)
So it's not the kind of thing you want to cut corners on. But it's also a lot of work and very expensive so of course you want to cut corners on it.
I suspect sometime in the next 5 to 10 years there will be a big economic crash that we Trace back to AI software used to do audits
Re: Also we are all very very soon get our (Score:2)
R u a bot?
Business pass efficiency dividends to owners (Score:2)
Fixed Cost Projects and Value per Unit of Time (Score:2)
A junior developer earns $10 an hour because it takes them 10 hours to solve a problem.
A senior developer bills $100 an hour because the same 10-hour project takes one hour. The productivity gain is reflected in higher value, not lower pay.
If AI reduces audit time, that is a measurable productivity gain. But time reduction alone does not define value.
If AI also improves detection, reduces risk, and increases reliability, then the value of the audit may increase even as the hours decrease.
Pricing purely on h
Just more hype (Score:2)
Surprised, anybody? (Score:2)