Latest / The Accounting Podcast / KPMG Withdraws AI Report Full of "Vibe Citations"

Transcript

KPMG Withdraws AI Report Full of "Vibe Citations" · The Accounting Podcast

• Attention: This is a machine-generated transcript. As such, there may be spelling, grammar, and accuracy errors throughout. Thank you for your understanding! David Leary: [00:00:04] State Farm this week announced that they're forcing all 19,000 plus agents to sign new contracts by 2027. So they're just ripping the band aid off, and they're going to tie their pay to AI outcomes, or they have to quit you either you either get on board or you have to quit coming to you weekly from the OnPay Recording Studio. Blake Oliver: [00:00:30] Hey everyone, and welcome back to the show. This is your weekly roundup of news in the profession. I'm Blake Oliver. David Leary: [00:00:37] And I'm David Leary. Blake Oliver: [00:00:38] And we've got our lead story today about KPMG's AI report that is full of vibe citations. That's a new phrase for me. We'll dig into that and more in this episode. But David, first, let's thank our sponsors. David Leary: [00:00:53] Here, our sponsors this week, we have Canopy Value Builder systems on pay and star search. Blake Oliver: [00:01:00] Let me ask you something. How much of your day is actually spent doing accounting? If you're like most firm owners, 30 to 40% of your time is eaten alive by the work around the work. We're talking, chasing client documents, drafting the exact same emails over and over again, manual filing, and trying to remember what a client said on a call last Tuesday. It's an administrative tax and it's killing your profitability. That's where canopy comes in. Canopy actually delivers on the all in one practice management promise. It handles everything from proposal to payment and all the steps in between smart client intake, tax workflows, month end, close automation and billing are all in one unified platform. No more duct taping ten different apps together. Plus, they have Canopy Coworker. It's a secure AI assistant that lives right inside the platform and actually does real work. It drafts context to where emails, summarizes client histories, take some eating notes and turns them into tasks automatically. Early access firms are already seeing what's possible when AI works inside your workflow instead of alongside it. To see what a truly modern automated practice looks like, head over to The Accounting Podcast dot promo slash canopy. That is The Accounting Podcast dot promo forward slash CANOPY, and that's a great lead in to our lead story about this KPMG report, David. We both spotted this and, uh, it's like another one, another, another Big Four report that is is full of citations that just are not real or are uh, uh, misidentified or all sorts of stuff. So what's going on here? David Leary: [00:02:43] And this has happened now so many times that I asked you if we already talked about this story, but no, it's another new story. It's another new incident. So KPMG just had to pull a major AI report after getting caught with completely fabricated case studies, all the victim of AI hallucinations. So big name organizations like UBS, NHS, Greater Manchester, Transport of London all claim that all the claims about their AI usage were completely false or misleading. Blake Oliver: [00:03:10] This is a 2025 report titled Total Experience Redefining Excellence in the Age of Agentic AI. David Leary: [00:03:19] And this is, you know, a month after we had this happen with E. And then I think Deloitte had some stuff maybe a month ago. It's just we have to create a database and just track these because the big four just keeps doing it over and over and over again. Um, and the ironic part of this whole thing is, you know, KPMG, just like Deloitte has an AI marketing page with an article on it that says essential elements of responsible AI, how solid guardrails can help you scale AI faster. Like they're, they're marketing themselves as the AI experts. The only way I could think this could work is now Big Four can go to fortune 500 and be like, look, we know all the mistakes that can be made. Now listen to us. Hire us because we keep screwing up so we know what to do or not to do. Blake Oliver: [00:04:01] This was detected by a platform called GPT zero, which is designed to detect whether text was written by a human or generated by an AI model, such as ChatGPT or Claude or Gemini, and it was originally developed to help educators maintain academic integrity. But it's grown into a broader toolkit used by journalists and enterprise teams and investigators to audit content for authenticity and factual accuracy. And they put out a report on their website about this KPMG report, and they found 45 citations that only five of which were accurate. 28 of the citations pointed loosely to real sources, but included altered or fabricated components. 12 citations were too vague or flawed to verify confidently, and at least 16 of those citations qualify as hallucinations under GPT Zero's methodology. It's hallucination check tool rated the report 89% flawed. The company also said that many of the report's underlying claims appear false, exaggerated or misattributed, especially where the report describes companies using Agentic AI and customer experience settings. So these are examples of Agentic AI that were hallucinated, and some of the examples are interesting. Kpmg described an Austrian utility, Verbund, as using AI agents for real time household energy optimization, including smart appliances and EV charging. But the cited source is a press release about Verbund venture arm, investing in a startup working on energy optimization. And so the report conflates the investor and the startup while adding unsupported details about agentic, AI and household level use cases. And there's just many, many more of these examples. The report incorrectly claimed that the airline Emirates, had a mobile chatbot named Sarah that could converse with passengers and change flights, but Sarah was actually introduced as a robot assistant in 2023 and did not have the flight change capability described. Another example is the report says that East Japan Railway was using AI agents for travel recommendations and disruption prediction back in 2019. But that's impossible, because this press release that was cited predates the commercial emergence of Agentic AI and doesn't mention Covid. David Leary: [00:06:30] I mean, that's before Covid even. Yeah. Blake Oliver: [00:06:31] Yeah. So, uh, another great case study. Another example of why you need a human in the loop and you need to verify, uh, what the AI is generating for you and not just push it out to clients or to the world. David Leary: [00:06:46] Well, especially if you're going to claim to be the expert on AI and train people. One of the things kind of related to this, I noticed, you know, how we, we, we shifted our spend on AI to cloud and we kind of shut down our subscription to open AI. Well, I've still been using open AI, but now I'm getting ads. Guess who my ads are from. Pwc offering me like AI expertise to implement AI. Like the these big firms are just like, you can't play off. You're the expert and then make mistakes like this. It's not going to work long term. Blake Oliver: [00:07:18] Since we're talking about AI, let's do a follow up to a previous story we've discussed that now includes AI. And this is the Pentagon's inability to pass an audit. Well, the Pentagon is turning to AI to help it do so ahead of the 2028 deadline that the government has set. They've awarded a nearly $49 million contract to Groundswell Corp. for an artificial intelligence based platform called Agentic Auditor. It runs through 2023. No. 2031. And it's going to help the Army, Navy, Air Force and other Defense Department agencies with data collection and audit preparation. The Defense Department is the only federal agency that has never received an unmodified financial audit, according to the Government Accountability Office. David Leary: [00:08:09] Did it say how much they're paying for this? Blake Oliver: [00:08:11] $49 million. David Leary: [00:08:13] 40 million. So basically, they went and signed another contract for $50 million to track their other millions and millions and millions of dollars of contracts. Blake Oliver: [00:08:21] Billions and billions and. David Leary: [00:08:22] Billions of. Blake Oliver: [00:08:22] Contracts. What is it? The Defense Department budget is now like a trillion. David Leary: [00:08:26] I think it's somewhere in there. Blake Oliver: [00:08:27] Yeah, something like that. So let's hope that it works. More AI news. Copilot co work is now generally available from Microsoft. This is well I haven't seen it myself but it appears to be Microsoft's version of cloud Co-work. So more than a chatbot, it can run complex, multi-step tasks from start to finish and give you the finished result, not just a draft. More than half of the fortune 500 have now used it since it was in a three month preview. And so if you have not yet used cloud Co-work because you're limited in what you can use in your firm or in your company, check out copilot Co-work. David Leary: [00:09:12] They're saying probably most accounting firms are in the office, 365 stack in the Microsoft stack. You probably have people in your firm using it right now. If you or you probably have it installed in your computer right now. Blake Oliver: [00:09:24] They also say it's cheaper than cloud Co-work. 30 to 40% cheaper on average per prompt. And that's based on Microsoft's own internal analysis of 125 test runs across 12 prompts. It's available in Microsoft 365 copilot user subscription license. So if you've got that, you should now have access to it. It's billed based on usage and the the price is going to depend for each task on the model, the context amount, the tool calls the runtime, that sort of thing. So this is great news because cloud has blown away the finance community, as we've talked about on previous episodes. And I've actually got an example that I'd love to share with you about some write up work that I did, some bookkeeping and accounting work that I did. David Leary: [00:10:19] And then I'm going to share with you a gripe about AI in accounting. So all right, you go first. Blake Oliver: [00:10:24] Okay. Well, you know, here's what happened is I needed to do two years of write up work, 2024 and 2025 for a, for a simple, you know, service based business. And this is something that in the past would have taken a lot of time, um, trying to go get C, S, V statements or CSVs from, you know, bank accounts online, right? Credit cards, you know, Amex, uh, a bunch of different bank accounts, some relay accounts, some not relay accounts. Sometimes you can get PDFs, sometimes you can get CSVs. You can't always get everything. So I got everything I could. I went in and I spent an hour or two downloading every single statement that I could find in whatever format I could get it. So it's a mix of PDFs and CSVs, and I dump them all into folders. And I pointed cloud cowork at the folders. And I wanted to do this right up in zero. And so I was working with cloud Cowork to figure out how could we do this most efficiently. And the plan I came up with was to get cloud Cowork, to extract all of the bank information from those CSV files and PDFs into a single file for each bank account for both years. So we're talking like nine accounts. And so there would be nine files. And these are going to be in a special format that Xero has to import. So column headers you know you've got to make sure everything lines up. David Leary: [00:11:54] Get everything FX compatible format. Blake Oliver: [00:11:56] Yeah. And you can what's neat with zero. I don't know if you can do this in QuickBooks is that you can do this. And then they have a special conversion toolkit website where you can connect it to zero, and you can upload the CSV and you can import all the transactions as reconciled bank transactions. So not only does it come in as a bank feed, but it comes in as a reconciled transaction if you include an account code. So I gave Co-work the chart of accounts, and I gave it access to all the bank statements in a folder. I just pointed it at the folder and I told it what I wanted to do, and I said, let's work on coding all of these transactions, categorizing them in Excel first. So I want you to make an Excel file for each of these accounts. And then we're going to work through coding them, you and me together. And then once we've got it all coded just how we want, we'll import it into Xero. And that ended up being a great approach. So Cowork pulled all of the transactions. It did the OCR and the bank statements where there weren't CSVs available. It merged all the CSV files together for each account, and it pulled every transaction for each bank account into a single file. And then it did it. And then it did its best to code the transactions. David Leary: [00:13:14] And that gives you a chance to do human review before it does the next part. Blake Oliver: [00:13:18] Yes. And one thing that helped is I took the, uh, GL detail from 2023, which I happen to have from the previous accounting system. And I gave that to Cowork. And I said, first, before you come back to me to code stuff, I want you to go through the GL detail, and I want you to do your best to categorize every transaction and then include a column in the Excel sheet for Confidence Score. So as a percentage, I want you to tell me how confident you are that this is coded properly. And that way when it created the Excel file, I could open that up and I could sort by that score, and I could look at the transactions that are less than 90%, for instance. And so I did that. And then I asked it to ask me questions about stuff it wasn't clear on. I gave it instructions on what to do. And basically within an hour or two I was it might have been less. Actually, I wasn't really keeping track. This whole thing took maybe four hours start to finish. And that included like gathering the statements. I was able to, uh, just give it instructions. It coded all the transactions. We didn't have anything in Uncategorized by the end, imported everything into zero via the import tool and done two years of write up work done in really what was a couple hours of my time. If you subtract the time to gather all the source documents. David Leary: [00:14:43] So it's scanning categorization and pumping out the the reports end to end. Blake Oliver: [00:14:49] Yes. And what's really interesting is that like, I was kind of worried that this approach would like miss transactions, and that's why I wanted to import it into Xero because I knew this is a simple business. It's just cash basis, right? I don't need to actually reconcile necessarily. If I'm confident all the transactions are in Excel, I could just get everything into a single sheet and then do a pivot table or something like that to get the PNL for the tax purposes. But I didn't want to do that because I wanted to actually verify that everything got in and reconcile. So I was able to do that when I went into zero and I ran the reconciliation reports, I confirmed that every transaction was in there via that method, Except there were only like less than six. There were six that got missed, and two of them got missed because there was like a statement date misalignment with the credit card. So only two transactions actually got missed legitimately and the others got missed because of a, of a weird thing with the import tool, where if a transaction has the same reference on the same date, it gets merged into like a single transaction when you import and, but, but what's amazing is that Claude was able to figure all this out for me. Generate two files to Reimport into zero to get all the missing transactions in. And we're talking 2200 transactions, so it missed six out of 2200. David Leary: [00:16:13] Yeah. Blake Oliver: [00:16:14] Immaterial really. And the amounts were immaterial too. Um, what's also really interesting is it helped me figure out a clearing account. So I had set up a clearing account for all the bank transfers. So thing about zero and I think this is with QuickBooks too, is you can't just import a transfer type transaction. If you want to do this quickly, you create a clearing account, which I call transfers clearing. And then you just code all those transfers in the import files into that clearing account. And at the end, hopefully it zeros out. David Leary: [00:16:47] Zeros. Yeah, that's how. Blake Oliver: [00:16:48] You but it didn't correctly. There was like a $5,000 balance in the transfers clearing. And normally this would be something where I now have to go through and like tediously try to figure out what wasn't a transfer. Well, Claude opened up zero in my web browser and looked at the detail and figured out exactly what was wrong. One was a returned payroll that needed to get coded to payroll expense, and accidentally got coded to transfers. It fixed that. And then the other was really interesting. It was there was an account that I didn't know existed. There was a business line of credit. And so there were transfers to like this Amex loan, this business line of credit that were in the transverse clearing account. It asked me for the statements. First of all, it suggested this might be the case. I went online into the Amex account, found the statements, dumped them all into the project folder, and it clod figured out that it needed to create a loan account. Ask me what to do. I confirmed it created an Amex business line of credit loan account, recoded all the transfers into that account, and then also broke out the principal and interest based on the statements for every single transfer. And at the end of this whole process, the transfers clearing cleared. Wow. And I had correct interest expense, and the loan balance was correct at the end of 2025. David Leary: [00:18:04] So you're doing all this with clod cowork? Yes. And so that would be an example of something somebody can do now if they're a Microsoft shop with the new Microsoft, what are they calling that thing? Blake Oliver: [00:18:13] Uh, copilot Cowork. David Leary: [00:18:15] Copilot Cowork. Okay. Blake Oliver: [00:18:16] Yep. So, you know, I, let's just think about this time savings here, right? So and going all the way back to when, you know, my work as a bookkeeper shifted because of cloud computing. So in the era of desktop, 2200 transactions would have taken days of work to manually enter into. David Leary: [00:18:36] Well, the real desktop, those those 7 or 8 differences, it takes you forever to find the last seven transactions and fix those. Blake Oliver: [00:18:42] That too. Well, if I enter them manually, it probably wouldn't have happened, right? But it would have taken days to enter all those transactions. Like, you know, I mean, you could calculate it like how many seconds per transaction and then, you know, 2200 transactions. We're talking days of work. Cloud would have reduced that to maybe like, I don't know, a lot less, but then I'd still have to import the transactions and then code them all in Xero using their tool. But now Claude's doing all the work coding, you know, 98% of those transactions. So we're talking something that used to take days is now down to basically a few hours of my time. David Leary: [00:19:22] For write up works A lot of times you kind of just need it good enough, right? Blake Oliver: [00:19:27] Honestly. David Leary: [00:19:28] Yeah. Because if people really cared, they wouldn't have waited two years to do their their bookkeeping or taxes. Right? So you probably just need to get it close enough. And yeah, this is a way more efficient. Now you could write because I've always been like a thing. I think people tried to attack at QuickBooks, launched a special version of QuickBooks QuickBooks ledger just to write up work. There's always been this, it never works, has never really fit in the cloud model because the cloud model is about billing for something month after month after month. And write up work is a one time thing. So it never really works, never really fit in the last decade of cloud accounting. So now you can really do write up work efficiently. Blake Oliver: [00:20:06] Yep. And I mean, I think it's probably if I ballparked it like a five times to ten times productivity boost, 500 transactions an hour, five over 500 transactions an hour, but. David Leary: [00:20:20] You can do data entry that fast. Blake Oliver: [00:20:21] Yeah. I mean, you know, it's it's like to data entry that would have been ten times slower. Um, perhaps I mean, it's hard to even say. It's, it's like there would have been so much pre-work to do all that using specialized tools to extract the data from the PDFs and then validating it and then importing it. I mean, and the fact that just cloud was able to troubleshoot this stuff for me and just chugging away in the web browser while I did other stuff is just incredible. David Leary: [00:20:48] So now when you do others, when you say this, right. Are you like using your same computer and you're actually doing this stuff in a different browser? Or is this like a siloed out computer that you're working like? Because don't both of you have to use the mouse as a mouse at the same time and type things? Blake Oliver: [00:21:02] Well, so cloud opens up a tab group in Chrome and I can just put that tab group in a separate screen. I got three screens here, and it just has access to that tab group. So I can keep doing whatever I'm doing on the other screens and just watch it work away in the background. And yeah, it's slow when it works in the web browser. So, um, one more step I forgot to mention is that after I imported everything in and, and had it all reconciled, I had it take a final pass at the GL detail and ask me if I wanted to recode anything. Like, like just look for inconsistencies, found 55 inconsistencies and asked me what I wanted to do. I said do this, do this, do this. Just in the, in the, in co-work. And then it went and it used zeros, uh, reclassify transactions tool, I think it's called find and recode. That's what it's called find and recode. And it went and did that manually. Like I would do it. It just did the binary code and that took the, did you. David Leary: [00:21:59] Tell it to go use that feature or did it just figure that on its own? Blake Oliver: [00:22:02] No, I figured that out on its own. Yeah. Wow. So opens up that tool, you know, does the filters, selects the transactions, records them. That was the slowest part just because it has to take a screenshot. Every time it does a click or a enters, it enters something with the keyboard. Um. Just imagine how fast it's going to be though, when zero enables transaction level access to cloud, which they're inevitably going to do because they've created this partnership with anthropic. David Leary: [00:22:28] And it's just you won't even see the mouse. It won't use the UI at all. It'll just, the data will just be fixed exactly appearing there. Blake Oliver: [00:22:33] So it's literally just like, you know, almost anything is going to be possible in both Xero and QuickBooks. I think pretty soon when they enable this kind of access, if they do it, I mean, we're not sure if Intuit will actually do that. You're skeptical, David, but even if they don't, they'll, they'll build their own like Agentic type of AI agent into the tool eventually, because they'll have to, because everyone's going to want this. So David, let's go ahead and thank our next sponsor. I'll let you read this one. This is value builder system. David Leary: [00:23:03] If you run an accounting firm, you've probably watched this happen. A client sells their business, they call wealth advisor, their attorney, their consultant. And you find out about the deal. When the K1 hits your desk, it's happening more than you think. 12% of business owners clients got written, got a written offer to buy their company within the last year. When that offer lands, it kicks off a wave of high margin advisory work, tax planning, estate work, succession planning, quality of earnings engagements that run often into six figures. But to most of your clients, you're the tax pro, so the work goes somewhere else. Value builder is an exit readiness platform built for accounting firms. We give you the tools and certification to start the end game conversation with your business owner clients before someone else does. In a typical 2 or 3 partner firm, our data shows a hidden million dollars in advisory fees sitting inside your existing client base. To see how Value Builder helps you move from five figure compliance fees to a six figure advisory engagements, head over to The Accounting Podcast dot promo value that is accounting podcast dot promo forward slash VALUE. Blake Oliver: [00:24:06] Welcome to our live stream viewers. Great to see all of you today. I wonder if you have the USA versus Australia World Cup. Game on. I'm missing it right now. And, uh, thank you for tuning in. Regardless of what's going on with the World Cup. Uh, we've got Dre here. Boring accountant. Great to see you as well. Three coffee emojis this morning. Jackson Pace Jackson says great seeing you guys at AICPA engage. Curious about your honest take on tax, GPT and some of the other tax prep apps go all in early or stick with tried and true processes currently using Proconnect. I've always wanted to keep my firm on the cutting edge of tech, but it's getting overwhelming at this speed. That is true. It's like the early days of cloud when there were so many different tools. David, I want to hear your thoughts. My initial reaction is go in on something so that you can learn it, but I'd avoid vendor lock in. David Leary: [00:25:07] Yeah, don't buy a five year contract or anything like, like try it out, use it with one client. Or like, you know, Blake and I had our company here. We run some companies on QuickBooks, some on Xero, some use anchor, some use ignition. Right? You like to put a client on different tools. It's not efficient for your firm, but long term it's efficient because you're going to know what tools do better than the other tools and it'll help you make the buy decision. So yeah, jump into experiment. But I was actually thinking about like these tax research tools, we were AICPA engaged. So you have what blue Jay, you have tax one. There's a bunch of these popping up and their argument is that, oh, we train them, especially around tax law. Yes. And my argument or point of view is long term. I think the base models are just going to do this. Blake Oliver: [00:25:56] If they have access though to those research libraries. David Leary: [00:26:00] Well, why wouldn't they? Right. Blake Oliver: [00:26:01] It's just well, because those are proprietary, right? Like the like the Wolters Kluwer Thomson Reuters have these like libraries of analysis and research. And if the if the base models have to like go through the tax code and the regulations and basically do that same analysis every time you ask a question, it's going to be either extremely expensive because they're having to do because. David Leary: [00:26:26] They're doing it from. Blake Oliver: [00:26:27] Scratch or yeah, PhD level research from scratch or it's going to be off because they're not going to be given enough tokens to do it. So that is the differentiator. David Leary: [00:26:40] I bet the base models will just do this out of the out of the gates automatically. It could be a year from now, but that's my bet, right? Like I would not lock into a what is it, blue Jays, $1,000 a seat for a tax research app that's locking into a contract. Seems crazy to me. Blake Oliver: [00:26:55] So stay monthly. Maybe keep your options open. Try different things. David Leary: [00:26:59] Stay monthly. Blake Oliver: [00:26:59] That's the good advice. Giles in the YouTube chat says, hi guys from Utah. This week Big Four have ingrained Four Eyes review processes. How do these errors get through that process? Well, maybe somebody is like too busy and just, you know, checks the box to say, I reviewed this and doesn't check all the citations. I think they'll get over this though, because they'll start running their own reports through these tools. That's what you need. You need both human review. And then you also need an AI that reviews for this type of hallucination. You can catch a lot of this stuff if you have the AI generating stuff. You need to have an AI that reviews it, but you also need to have the humans as well. David Leary: [00:27:41] Yeah. And that pausing along the way or having those facts like the pause, right. And the human involved and well. Blake Oliver: [00:27:47] And that's why when I did this write up project, I didn't just say, here's the GL detail from 2023. Here's all the transactions from 2024 and 25. Go encode them all and enter them into zero. I wanted to review it first and I caught significant stuff and got it correct before. David Leary: [00:28:06] I was calendar appointments. I'm like, hey, put this on my calendar before you do it, just give me a table of the events so I can see it. And then yeah, it shows me, okay, it interpreted correctly, it got into a table correctly. In theory, it's going to get to my calendar correctly. It could still hose up. Blake Oliver: [00:28:19] But and the key. David Leary: [00:28:20] Is. Blake Oliver: [00:28:21] I didn't have to review everything. I just reviewed the exceptions. Anything that was below 90% confidence is what I reviewed. And I could have set that higher. And actually, maybe, you know, I'm trying to think I probably did. I did look at everything, but I looked really closely at the transactions that were low confidence. David Leary: [00:28:37] All right. Can I, can you put your gripe department hat on now? I have a. Blake Oliver: [00:28:41] Great tell me. David Leary: [00:28:42] About you. Blake Oliver: [00:28:42] Go for. David Leary: [00:28:43] It. We're now that your mark, we're big enough. We're starting to have intercompany transfers. So Blake I'm like, hey, give me the bills. Send me a bill a month. So he creates a bill a month, but he puts them all into one PDF. I'm like, ah, all right, I know how this is going to go. So I took them, I import, I put that PDF into QuickBooks. What did QuickBooks do? The AI and QuickBooks just makes one huge bill with a line item from each of the other bills. Blake Oliver: [00:29:08] Uh, right. David Leary: [00:29:09] Now, it's just a big mess, right? I like technically we can scale with the money moved around, but you know, we're anal, like we're accountants. I want to have a bill and a payment lined up accurately based on the date of each bill, blah, blah, blah. Well, I started thinking I was really thinking about this because I'm doing a lot of vibe coding. The thing that these AI agents, Jax doesn't have, uh, intuitive cyst doesn't have, uh, from what I've seen of Oracle and nets or sages AI things that nobody has what vibe coding has. So I've been doing a lot of vibe coding and vibe coding has two, two modes, kind of a, a chat mode and then a build mode. So you chat or you're an ask and you start doing things and then it comes back and it says, hey, I noticed in your database you have this field, would you like A or B? And then you answer A and then it clarifies. It says, okay, here's my plan. I'm going to do this. Switch me to build mode so I can go do it. David Leary: [00:30:03] Like why these agents in these accounting systems do not do the same thing. I should upload this. It should say, hey, I noticed you uploaded this bill. There's 25 bills in here. Do you want one bill or do you want 25 separate bills? And I would just answer then, then would switch to go do work mode. Like they're missing something that all the vibe coding products have, the AI coding tools have. And maybe it's a learning curve, I'm sure, because in the beginning, the coding tools tools didn't do this. They just started writing code and you were like, stop, pause. Right. And so we have to get this into accounting because it should just all I have to do is clarify with me. But no, now I got to take this. I have to delete the transaction from QuickBooks, take the PDF into some other AI, and I'm going to create 12 separate puffs from this PDF. Then I'll reimport it in to get where I want to go. But if it just asked, like there has to be an ask, that has to happen along the way. Blake Oliver: [00:30:52] David. I'm going to take us in a different direction after our next ad and that is from on pay. Are you tired of payroll headaches getting in the way of the client experience you want to deliver? Manual workflows creating bottlenecks, compliance, nightmares, and endless support calls that go nowhere. There's a better way for your team and your clients on pay. Is the payroll partner that accountants and bookkeepers actually love? Why? Because it's easy to use, packed with value, and backed by support that actually supports you. Their team gets rave reviews for being fast, expert, and actually reachable when you need them on pay. Handles the heavy lifting. You get a dedicated onboarding coordinator who sets up worker profiles and transfers year to date data from previous providers, all at no extra cost. Their seamless QuickBooks and Xero integrations eliminate manual journal entries, and they support any type of business you serve farms, restaurants, nonprofits, you name it. On pay can handle unique requirements without adding complexity. And on pay keeps pricing simple, too. Everything your clients expect from multi-state filing to off cycle pay runs. It's included no hidden fees and no surprises. To book a demo. Head over to The Accounting Podcast dot promo slash. That's The Accounting Podcast dot promo forward slash ONPAY. Let's talk about pricing in CPA firms. You would think that with all of these AI tax prep apps, that tax might be getting compressed when it comes to pricing, if it gets easier to do, it should probably get cheaper as well. Right. Well, that's not the case at CPA firms. Cpa Trendlines is reporting that accounting firms are recovering some of their ability to raise fees after a week period in 2024 and 2025, the strongest pricing momentum is in tax prep, tax planning and advisory services, and audit pricing remains comparatively soft. Overall, CPA firm pricing is up 4.2% year over year. That reverses a 2.1% decline from the prior year, and tax prep and planning show the sharpest increase at nearly 8%. Advisory focused work went up over 6%, audit only, up 2.3%. Bookkeeping and compilation sit in the middle with about a 3% growth rate. David Leary: [00:33:16] So it seems general, it's all just inflation. If you start averaging all this out, like accountants just are still keeping inflation, but at least they're not behind inflation, which I think was true two years ago. Blake Oliver: [00:33:27] Well, I think the important thing to look at is the, you know, relative price increases. I mean, you know, 8% is a lot bigger than 2%. Yes. On a, on a relative basis. And so I think it's good news. It means that clients are willing to pay for tax planning advisory for the human in the loop to make sure that the numbers are right. And AI is not putting pressure on those fees at this point. And I, I don't expect it to. David. I'll let you decide where to go next. David Leary: [00:33:59] Um, I want to talk about this, uh, State Farm AI plan. Uh, so State Farm, 100 year old company and their business models essentially have thousands of human insurance agents. You probably have an insurance agent, right? Blake Oliver: [00:34:13] I have State Farm, I have. David Leary: [00:34:15] A farm, and you probably talk to the human, and the human asked you questions, typed it into a computer. You went face to face. You probably met them, typed it in a computer, then they gave you a quote, and then they just take the money out of your bank account every month, and they get a commission every year for the rest of your term that you're you're in a contract with them. Well, State Farm this week announced that they're forcing all 19,000 plus agents to sign new contracts by 2027. So they're just ripping the band aid off, and they're going to tie their pay to AI outcomes, or they have to quit you either you either get on board or you have to quit. Um, they're rolling out a new AI agent called Na'vi, and consolidating offices and slashing agent commissions by up to 40%. Now, the way they did this was not great either. They brought all these agents to Los. Was it the conference in Las Vegas? I'm just I'm having a click problem here. What did I just do? Blake Oliver: [00:35:09] They brought out all their agents, too. David Leary: [00:35:11] Yeah, they brought them all out to Las Vegas to a big conference. They had pink a pink concert. Jimmy Fallon was there. And that's where they explained that this change is happening to all these agents. Now, this is like major disruption because of AI. I know we've talked about like, where are the jobs going away? But the reason they have to do this is because they have competitors now that are selling insurance, and they're growing by 300 or 400,000 people a quarter. And if you really think about it, you could just buy insurance from a chat bot. What's the difference of going face to face and sitting at their desk or answering questions on a chat bot. Blake Oliver: [00:35:46] For simple policies? It really is just an information gathering exercise. That's my experience anyway. Like if you're buying homeowner's insurance or you're buying car insurance or, you know, not something super complex, you talk to that agent on the phone or via email and you're giving information and they're just entering it into a system and then it goes into underwriting. So this is something that, yes, like an AI agent could totally do. I think the area where they're not going to be able to automate this is with the more complex policies where there's a lot going on. But yeah, I guess what's the takeaway for accounting? It's that if, if, if your job is taking information in putting it into a system, and that's mainly what you're doing in addition to the customer relationship, like a good chunk of that is going to be automated. Ai agents can do that. David Leary: [00:36:41] And it's not so much the like this concept of AI is going to take my job. It's, it's another startup that's going to use AI to build something for the consumer that's going to take away your job, right? It's just helping other companies move faster and offer the same services. And that's what this is. It's services. Are you offering services fast enough and at the right price point to compete in the future market? And this this commissioned model of 19,000 people just is not going to work in the new world. Blake Oliver: [00:37:10] Space went public. We talked about that in a prior episode. I'm a little bit I was a little bit skeptical, given that this IPO combined a bunch of companies that must controls into basically a single offering, right? We've got not just space, we've got Starlink, and then we've got all these AI bets all in one place. David Leary: [00:37:32] And Twitter, right? Blake Oliver: [00:37:33] And oh yeah, Twitter to write X. Uh, and didn't seem to deter investors. The stock popped something like 50% in the first few days and it's come back down. I think it's now up. Let's take a look at the share price. It's up 14% 15% as we record this on June 19th. So, you know, anyone who bought in has done pretty well if they've held I mean, that's not bad. Five days, a few days and you're up 20%. Um, and there's a accounting story tied to this. Well, it's the CFO of SpaceX who was featured in cfo.com. His name is Brett Johnson, and he is now a billionaire as a result of the IPO of SpaceX. That's amazing. You don't you don't hear a lot of you don't hear a lot about a lot of CFOs who become billionaires as a result of their company going public. And it's because Johnson took a lot of equity in, in, in as compensation instead of salary. Um, let's see what the numbers are. So in 2025, Johnson received total compensation of $9.8 million, which was an $825,000 salary and $9 million in stock awards. So total comp of 9.8 million. Um, you know, compare that to the Tesla CFO who got like, what is it over? I had that in front of me a lot more, a lot more in cash. So I want to say it was like 100 could have been 100 million in cash. Does make that might not make any sense. But, um, yeah, CFO now a billionaire. So when for the accountants there. Yeah. David Leary: [00:39:31] So I have a related story to this space. Um, going public. Historically speaking, Blake, you as an accountant or our listeners, accountants, what are the quote unquote traditional financial disclosure channels. Blake Oliver: [00:39:44] The traditional financial disclosure channels. David Leary: [00:39:46] Yeah. So every quarter I'm going to run my quarterly reports. Like how do I normally distribute that? Blake Oliver: [00:39:51] Um, like financial, uh, investor media, you know? Yeah. Like Bloomberg and, and all these other sites. David Leary: [00:40:00] Or I have a site on my page that's investor relations and I have the slide deck and I put the color and then it gets picked up or there's a press release or whatever. Well, Elon Musk is going to ditch this traditional disclosure channel method. And he's only he's only going to release, uh, SpaceX's quarterly and annual results through their own website and X and that's it. So X slash Twitter. So they're not going to utilize the regular traditional investor relations media companies that exist. I just, I'm trying to understand like, like, is it like, why do this or like, or what other people's opinions are of this? Or is this also just much to do about nothing? Because at the end of the day, if I think about Intuit, they have an investor's page and everything that's there is the same stuff that gets sucked into press releases every time they do. Like, does this even matter? Is this even a story? I don't know, why would he even say this, right? It seems odd. Blake Oliver: [00:40:57] To drive people to X, perhaps. Hey. David Leary: [00:41:01] I could see that. Maybe. Blake Oliver: [00:41:02] Okay, I've got a related story. Since we're talking about, uh, public companies we've covered in the past this, uh, move by the SEC to allow companies to report semiannually twice a year instead of quarterly. And the CFA Institute did a survey that found that investors broadly oppose the idea to allow companies to shift from mandatory quarterly reporting to semiannual. According to the poll, investors believe quarterly reporting means main remains critical for transparency, comparability, and market efficiency. So what percentage are we talking about? It's 62% of respondents oppose replacing quarterly reporting with semiannual. 63% said the benefits of quarterly reporting outweigh the costs. Um, so that's the numbers. Now, I was curious who exactly got surveyed in this report. And so I dumped the report into Claude and asked it to tell me who got surveyed. It was a, um, it was exclusively investment professionals. So it was a random sample of CFA charterholders globally employed as either investment analysts or portfolio managers. And they got, you know, a few thousand of them to respond. So what is not included in here is retail investors. And that is what I'd be interested to know is like, do like retail investors care, would they be fine with semiannual? And I think they I think they generally would be like, that's my feeling without knowing it, because I feel like retail investors don't actually look at the financials anymore. David Leary: [00:42:45] No, I had another article and I wound up not bringing it, but it was like retail investors invest using FOMO. That's it. That's all they invest. They don't look at financials. They just invest. They they're afraid they're going to miss out. And that's what makes them invest. Um, I, I just, I feel like that this creates pressure on companies to hit numbers every quarter. And some of the arguments, I think in this argument, they, uh, they said that there's concerns about, you know, how often information like maybe bad news doesn't get disclosed as frequently, maybe it should. There's I could see those kind of arguments for quarterly reporting, but I just, I feel like it just creates weird, weird, uh, weird games, right by CEOs, CFOs to hit their numbers. Because basically, instead of focusing on their customers and their company and their product, they focus on the next quarter's release and you. I got to basically the streets, the customer, not the, you know, the customers. Blake Oliver: [00:43:43] Given who was surveyed, I feel like actually 62% supporting keeping the quarterly reporting is actually really low. Like it should be a lot higher. If quarterly reporting really mattered, that number would be bigger. David Leary: [00:43:57] Yeah, that's a good way to, to, to think about this. If, if it was so important, it would be like 90% oppose it. Yeah. Blake Oliver: [00:44:06] Shifting gears again, a new global study reported in accounting today says that freelance finance and accounting has the widest gender pay gap of any major freelance work category. So freelance accountants, females versus males, widest gender pay gap of any major category of freelancers. Here's the number. Women in these roles charge 26% less per hour than men, on average 26%. So freelance finance. Accounting. Dollar wise, it's. Women are charging an average of $30.77 per hour. And then compare that with $41.61 per hour for men. The largest gap appears in accounts payable management. Women are earning 35% less than men. Bookkeepers. Women are charging 24% less, management consultants 23% less accountants 22% less. What's interesting is the the only area that is close to parity is fundraising consultants, where women actually charge 0.4% more than men on average. So fundraising consultants in the nonprofit world, it's about equal. David Leary: [00:45:27] From a behavior standpoint, this doesn't make sense because in theory, we're financial professionals. We should understand the numbers better and this should not happen. But if I'm if I'm understanding this study correctly, uh, wedding photographer, a freelance wedding photographer who's male and a freelance wedding photographer who's female, you're going to get the same price from both of those. Blake Oliver: [00:45:44] Well, I don't know about that categories, but I just know about this one. This is the. David Leary: [00:45:49] They said that we're worse than other industries, right? Other freelancers. Blake Oliver: [00:45:52] There were like ten other major ones that were analyzed, but not necessarily all. Yeah. David Leary: [00:45:56] So imagine it'd be like a wedding photographer type situation. But when it comes to accounting work, we just. Blake Oliver: [00:46:03] We have the greatest disparity. David Leary: [00:46:05] That does that. That seems crazy because we're the ones that should be good at the numbers. I don't that's how does this happen? Is it just historical, you know, white male in charge of the accounting world for decades, hundreds of years. Like, is that like, how does this happen? Blake Oliver: [00:46:21] I don't know, I'd be curious if any of our listeners, live stream viewers have any ideas. We will be watching the chat. All right, here's a here's a fraud story. We don't cover fraud very often on this show. And this one caught my attention just because of the name. It's kind of wild. And it was reported by TMZ. So we have TMZ covering some fraud here. Um, according to a federal indictment described by TMZ, Frank Mazzochi, a former account manager at a Beverly Hills management and tax firm, used his role handling, handling celebrity finances to gain access to Nick Cannon's accounts. So, Nick cannon, does that ring a bell? David Leary: [00:47:10] David, is he the guy that has like 17 kids? Like he can't afford to have money stolen from him? I think. Is he the host of that, uh, America's Got Talent or something like that? Blake Oliver: [00:47:19] Well, he started out as a teenager, uh, on television on all that. And then he went to go on to host shows like wild and Out. America's Got Talent and The Masked Singer. So he's done really well for himself. And prosecutors are alleging that his accountant controlled his debit cards and pins and used that access from 2019 through 2023 to take money without Nick Cannon's knowledge or approval. And it was this is wild one about $1.7 million in ATM withdrawals, $1.7 million in ATM withdrawals in a four year period. David Leary: [00:48:01] That's a lot ATM withdrawals. Yeah, because, like, most ATMs don't have that kind of cash. Blake Oliver: [00:48:05] I mean, how often was he taking cash out? Right? More than 165,000 in Amazon purchases, nearly 192,000 in personal travel and more than 160,000 in other personal expenses. Yeah, it was over $2 million. He was charged with five counts of wire fraud, three tax evasion charges, and he concealed more than 1.7 million in stolen income on federal tax returns for 2021 and 2023 and 2022, and the firm is not identified in the indictment. The firm terminated Mussorgsky after discovering the missing funds in July of 2023, and authorities believe that he has now fled to Uganda. He holds a dual US Ugandan citizenship, it sounds like. David Leary: [00:49:04] What is a celebrity or an athlete supposed to do? Who has massive or musician that has massive amount of money and wealth happening? And this is all going on because the person that has the expertise that in theory would help you prevent this from happening is the person committing the crime. Did you should you always. Would the advice be, go get two accountants that don't know each other. Blake Oliver: [00:49:24] Have one. Check the other's work. Yeah. Separation of controls. I mean, yeah, I mean, like, this is the problem with business management is you have the same firm, uh, has the ability to move money as is doing the books. And so all it takes is poor internal controls of that firm. One accountant who is, is reconciling the accounts and also has access to all these cards, which happens all the time. And the partners may have no clue this is happening. And the client may have no clue this is happening. So how do you prevent it? I think actually that's a great idea. David. Have two different firms, have one firm, do the books and the other firm move the money or act as like the CFO? Right? David Leary: [00:50:07] Yeah. Every celebrity athlete, uh, musician should hire two accountants. Blake Oliver: [00:50:12] And and then, you know, I mean, there's also like standard stuff like, uh, this is the recommendation for business owners all the time is like, have paper copies of bank statements sent to your home address like duplicates, right? Don't don't have them sent to your accountant's office so that you can just like periodically go through them. And I mean, imagine if all you did was look at the bank statement. David Leary: [00:50:33] Tell that person that that's a good idea to do, right? In theory, it would be their accountant and their business advisor who's not going to tell them to do that. Blake Oliver: [00:50:40] So maybe your suggestion is how do they get to accountants? Yeah. David Leary: [00:50:44] Yeah. Blake Oliver: [00:50:45] Um. All right. Let's see. Where do we go next from here? We're just moving around all over the place today. David, let's go back to the big four. Uh, KPMG in the UK has killed its summer perk that let staff leave work at 3 p.m. on Fridays. Nearly 17,000 UK employees will now have to stick around a few hours longer between June and August. The firm says it is about, quote, market conditions and business needs, but there's an extra two hours on a Friday afternoon really make a difference. I mean, I wonder. David Leary: [00:51:20] If. Blake Oliver: [00:51:21] This is the old is this the old mentality of like, the partners are saying we need more charge hours, so we're going to keep the staff later on Fridays during the summer so we can get our charge hours up and we can make our numbers. This is the problem with that whole billable hour mentality. David Leary: [00:51:39] It's the it's the mindset. Blake Oliver: [00:51:42] Yep. Um, KPMG has already laid off. David Leary: [00:51:44] Hundreds of workers. This year is probably a great. It was probably a great benefit that motivated people to try to rush to get their work done before. It's because they want to get out that two hours early. And so now if they're going to be stuck there for two hours, they're probably going to, in general do less work in total because they're not getting motivated. Blake Oliver: [00:52:01] I feel so sorry for those accountants in the UK. I mean, you know, like the weather is good there only a few months out of the year. So it's. David Leary: [00:52:08] Like. Blake Oliver: [00:52:08] This is your window. It's like a couple hours outside in the sun. And, uh, now you got to stick in the office until five. David Leary: [00:52:14] And meantime, the partners are taking six hours off on Fridays. Blake Oliver: [00:52:17] Right to go golf. I don't know what they do there. Okay, David, you want to take the next one? David Leary: [00:52:25] Yeah. I have some Intuit news, which I thought was kind of interesting. Okay, so the headline was Intuit Enlists Mother NY to enhance QuickBooks. Like, what does mother and why? Mother and marketing a marketing creative partner. So the move is about making QuickBooks feel more human and emotionally resonant as AI powered accounting software gets increasingly crowded and competitive. Mother's job is to position Intuit's AI platform as a confidence builder for small business owners across four major points. So essentially, they're trying to humanize AI. And so it made me immediately wonder, like, will they be starting with the CEO, Suzanne? Because Suzanne, when he talks to accountants, he comes off very AI and robotic. Like he doesn't come off very human. Like you have to start from the top, right? You can't just do commercials like this or change in the product. Like you have to start with like making Intuit become human again, because Intuit kind of got away from being human. I would argue. Blake Oliver: [00:53:18] Well, remember when they had those giant robots, uh, ads at the Super Bowl or something like that? Like that was. David Leary: [00:53:23] That was that was the first taste of this was coming. Yeah. That's true. Blake Oliver: [00:53:28] Let's do some follow up about tariffs, everyone's favorite topic. So, um, the administration's tariffs that were implemented under the International Emergency Economic Powers Act were overturned by a court order. And so it was looking like these tariffs were going to get refunded. And this was the US Court of International Trade. They'd ordered Customs and Border Protection to refund about $166 billion, plus interest from tariffs collected in 2025 and early 2026. And the administration was looking like it was going to go along with it. But now they are challenging that court order. And so, um, It's complicated because the Customs and Border Patrol started issuing refunds and the because the administration didn't immediately appeal. And so we we assumed that the government was going to comply. But now they've reversed course. And they are arguing that the court order functioned like an improper nationwide injunction and that the CPB cannot refund liquidated claims unless there is a company specific court order. And under that interpretation, businesses with liquidated entries would have to file their own lawsuits to recover tariff payments and interest. So all these companies would then have to file their own individual lawsuits. And so basically, about 81 billion, nearly half of those duties paid has already been liquidated. Cpb had refunded about 20% of the estimated 166 billion. And so they're basically trying to hold on to some of that money. Like a good chunk of it looks like half of it or something like that. So the tariff uncertainty continues. Let's also talk about some more administration follow up. This is the IRS and Ice a new taxpayer sorry, not taxpayer. A new Treasury Inspector General for Tax Administration report found that the IRS provided last known address information for about 47,000 individuals to Immigrations and Customs Enforcement. The disclosures came from an April 2025 data sharing memorandum of understanding between Treasury on behalf of the IRS and the Department of Homeland Security on behalf of Ice. And so now we know the number. David Leary: [00:56:01] And this is the first time I've seen the actual what data was shared? Yes, because I've been following this story for a while, and I haven't seen what data they shared. So it basically addresses physical addresses. Blake Oliver: [00:56:13] This is controversial because federal tax information is generally protected under IRC section 6103, which limits disclosure unless specifically authorized by law or by the taxpayers. And that's why taxpayers generally expect that information they provide to the IRS will remain confidential, absent legal authorization. So there are all these questions about the legality of this arrangement. David Leary: [00:56:37] And this is my argument of the Trump tax returns being leaked in the Trump lawsuit. Like this could be a class action, massive class action because like, what's the price? Is it $1 billion per return now? Like, like how much precedent has been set here because firms have to pay. You have significant penalties, right, Blake. As a CPA, if you leak taxpayer data. Blake Oliver: [00:56:58] Oh, yeah. It's a big deal. David Leary: [00:57:00] Yeah. So there's got to be lawsuits that are going to happen. Class actions, massive class actions, more than $1 billion. Blake Oliver: [00:57:08] So the Inspector general also reported that before releasing the data, the IRS created an automated matching process to compare Ice records against IRS records. But they found that the criteria for matching did not identify records accurately or consistently. So of the 1.2 million records that Ice requested, address information for IRS ultimately supplied that 47,000 addresses people, and they rejected the records that didn't meet their conditions. But the inspector general said that the IRS process failed to screen out all records that should have been rejected. So there was no response from an IRS official included in the report. Um, and yeah, that's the news there. All right. David, I think we are at time for this episode. Dear listeners, don't forget that you can earn free continuing professional education for listening to this episode and all our past episodes of the Accounting Podcast on the earmark app, go to earmark.app in your web browser, or get the free app from the App Store or Google Play Store. Create a free account and you can earn one free CPE per week. Sign up and for a low price of $170. As we record this, you can earn unlimited CPE for the whole year. That's going up to $200 in July, so sign up now and lock in that $170 price. David Leary: [00:58:37] By the time you hear it, you have about seven days to lock it in. Blake Oliver: [00:58:40] And it's not just our podcast. You can listen to shows like oh My Fraud and Federal Tax Updates and the unofficial QuickBooks accountants podcast. And David, any others in our top ten? Many, many more. And you're going on, you're going on to earmark to check it out right now. David Leary: [00:58:58] I got a whole website, The Accounting Podcast dot com, and I can easily see all the ones that are in earmark ranked. So we have ones with the channel. So we have 4040. The top 800 accounting podcasts all have channels in the earmark app. So you can do federal tax updates tax inaction. You know that's a great way she counts up who's really the boss slash tax. Mr.. Our show audit smarter best metrics. I mean there's just a lot of them. Check it out. Blake Oliver: [00:59:28] Earmark.app. Thanks to everyone who joined us live. You can check us out on YouTube, search for The Accounting Podcast, subscribe to our channel and hit that notification bell icon to get notified when we go live and join us as we stream. You can chat with us. Thanks to everyone who joined us. Great to see you, Penny Breslin. Um, so many more. Uh, we got Heather Smith here. Uh, Tate also great to see you as well. Thanks everyone who commented. See you around here next week. Thanks, David. Bye, everyone.