Could AI Replace CFOs & Time Tracking 188 IRS Calls

0:04David Leary: A North Carolina woman, she lost$600 ,000 in a romance scam. The IRS then sent her a bill for$225 ,000 in taxes and penalties. Why did that happen? Because the money she sent the scammer came out of her retirement account, so it counted as taxable income.

0:23Blake Oliver: Coming to you weekly from the OnPay Recording Studio.

0:29David Leary: Hey, everyone, and welcome back to the podcast, your weekly roundup of news in the accounting profession. I'm Blake Oliver.

0:36Blake Oliver: I'm David Leary, and I'm BeatBlake. I got that shingles vaccine this week, and it's hit me. So I'm going to push through, though, for the show, for the listeners. We got to push through here.

0:45David Leary: The show must go on. Today, we're talking about whether or not AI could replace CFOs and other top executives, some new survey data on what U.S. adults think about that. And an EA time tracked 188 IRS calls and reported the results. And we've got them for you. All that and more in this episode of the accounting podcast. But first, David, let's thank our sponsors.

1:10Blake Oliver: Our sponsors this week, we have OnPay, Numeral, Savant Labs, and Cloud Accountant Staffing. Are you tired of payroll headaches getting in the way of your client experience that 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. OnPay 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.

1:44Blake Oliver: OnPay handles all the heavy lifting. You get a dedicated onboarding support specialist who sets up worker profiles, 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 businesses you serve, farms, restaurants, nonprofits, you name it. OnPay can even handle unique requirements without adding complexity. And OnPay keeps pricing simple too, everything your clients expect from multi-state filing to off-cycle pay runs. It's included, no hidden fees, no surprises. To book a demo, head over to accountingpodcast.promo.com.

2:20Blake Oliver: That is accountingpodcast.promo.com.

2:24David Leary: Thank you, OnPay, so much for supporting us over the years and being our lead sponsor. Please do use those links that lets them know that you found them through the podcast. They appreciate knowing where you came from, and it helps us get more sponsors. All right, David, let's talk about whether or not AI can replace CFOs and other top executives. This is something I've wondered about myself because I'm automating more and more of my job as a CEO, as a founder. And like a lot of it, just following up on stuff, pushing stuff forward, a lot of it is automatable. And you've seen that. You've been getting these emails from me and comments and all this stuff.

3:05And this is a smart org where you're directly, all the reports kind of roll back into you, right?

3:11Blake Oliver: So I think there's more testing. But if you think about it, like at a bigger corporation, the CEO has this team of people that are just building slide decks for them. So I don't know if the CEO could be replaced by AI, but that next layer of direct staff under the CEO, yes.

3:31David Leary: Maybe people who are mostly just doing project management, executing that CEO's vision, perhaps. Let's see what the survey data says about how Americans feel about this. Watch Chest surveyed over 2 ,000 US adults and asked them whether AI should replace CEOs, CFOs, and other senior executives. And a lot of Americans are in favor of this. There's a steep gender divide though. So get this, it's 45 % of men who favor replacement. 45 % of men say, I would be happy if they replaced these executives with AI.

4:09Blake Oliver: Men hate their bosses. Okay.

4:11David Leary: But only 25 % of women said that. I mean, that's a 20 percentage point difference, right? Quarter of women, almost half of men. The survey gatherer, WatchChest, this company that did the survey, they suggested that women may be more reluctant to remove human judgment from decisions involving jobs, ethics, culture, employee well-being, privacy, and accountability. support also changed depending on how much money you make. So people earning$150 ,000 to$200 ,000 a year supported this the most, the upper middle income spectrum, upper middle class. And that peaked at 55%. So over half of those earning$150 ,000 to$200 ,000 a year supported replacing executives with AI.

5:00David Leary: I can have a theory on that. Tell me why.

5:03Blake Oliver: The average bottom of the rung employees and possibly most of middle America has never talked to the CEO, has never met the CEO, and definitely has never worked with the CEO or been in a meeting with the CEO. But people that are high income probably have higher level management positions and they either think the CEO is an idiot, they don't trust the CEO, they're like, our team does all the work for the CEO. So the income is probably just based on people's experience with CEOs.

5:29David Leary: And maybe at that income level, you're sort of in the middle of the corporate hierarchy where you're getting all these ridiculous demands from on top and you are supposed to somehow execute these crazy visions.

5:42Blake Oliver: You're changing the shade of yellow in a slide deck.

5:44David Leary: Yeah, there's bad executives, right? And you are at the mercy of them. And then you have to implement that and tell that to all the people that work for you. I mean, if you think that management at your company is dysfunctional, which it can often be at a big company, then yeah, I mean, you would support it, right? So I was thinking about this, right? Like, let's think about this from the CFO perspective. We are seeing AI absorbing a lot of the work that finance teams do. It's taken over with forecasting, monitoring, month-end close workflows, strategy support, M &A type analysis, all of that.

6:26I mean, AI is, everybody is using it.

6:29David Leary: If you are in a role where you have to do complicated spreadsheet analysis and you're not using like Cloud Cowork by now, I mean, you're a dinosaur at this point. If your company allows it. And so many people are using it, even if their company doesn't allow it. I think we talked about on the show a week or two ago.

6:45Blake Oliver: Shadow AI use, what's that called?

6:47David Leary: Shadow AI use is like a third of employees. And so, you know, I mean, but let's tackle the question of whether or not AI could actually replace these leaders. And my take is definitely no, not at this point, because the thing you have to do to make it work right now is you have to really monitor it. And you have to correct it. You have to read all this analysis that it gives you, and then you have to tell it where to go. And if you just let it operate autonomously, you get terrible results at this point. So I am spending a ton of my time now just getting through a bunch of simultaneous chats that I'm going.

7:30David Leary: I might have 20 going at a time and reading what it gave me and then telling it what to do next, which actually is hard. It's exhausting.

7:40Blake Oliver: Oh, yeah. I'm saying, but I have a couple AI agents doing different things. And yeah, you're constantly looking back where you're at, reviewing. It's a lot of context switching, switching more than ever. Like the, again, we talked about this months ago, that cognitive load. Even though more is getting done and you feel more productive, it's not, it's still a lot of work. Like you're not actually doing less work. You're actually working your ass off, but it's just going so much faster. And that's why it's exhausting.

8:06David Leary: You're working your brain instead of sitting in meetings and then delegating in meetings all day long, which if I have to choose, I'm absolutely going to choose the current system of AI doing this project management for me and me directing it. It's so much better. I don't have to sit around and wait, but I have to have so many chats going because I can't just sit there and wait for it to chug away. I mean, I get bored instantly. I'm just sitting there while it goes. And it takes a while to get back to you sometimes, right? Minutes. And I just, so then I have too many going. And then, like you said, I'm context switching and it gets mentally tiring, mentally exhausting.

8:46David Leary: And that's one of the reasons why I think we're going to end up working a lot. Well, you know, whether we work fewer hours is debatable, but I don't think it's possible to be like doing what I'm doing and work more than six solid hours a day. Like I can't do it. I am exhausted at the end from the mental load. The hours work that we may do is gonna still feel like 40,

9:09Blake Oliver: but we might only be working 18. It's just gonna feel like 40.

9:12David Leary: Exactly, because you're doing heavier reps with your brain, right? Instead of just sitting in meetings all day, which is a different kind of endurance. Exhaustion, yeah. Yeah, so that's my take. I should say I spotted this thanks to CFO.com. They covered this on their site. Oh, I should also add the regions mattered. This is interesting. The West Coast supported this the most at 42%, followed by the Northeast at 38%, the South at 30%, and the Midwest way behind at 18%. So Midwesterners don't think this is a good idea, don't want this. But Californians, Arizonans, Washingtonians, Oregonians, I guess, you know, they like it.

9:58David Leary: So I guess the question too is, and a great follow-up would be, does this say that people trust AI to run a company or does it just indicate how much they don't like the humans who are running the companies now?

10:09Blake Oliver: I think it's a dissatisfaction survey. Yeah.

10:11David Leary: But it would be interesting to somehow run a test of this, is create a company, a business, and put an AI in charge as the executive team, maybe a team of AI agents, and see what happens versus a team of humans. And I think, of course, you're going to get widely divergent results because AI trends toward the middle. But honestly, an average executive team could still generate profits for investors. So it might be a strategy.

10:45Blake Oliver: Many times the senior leadership team and the CEO's admin basically run the CEO. They're the ones running the CEO. So essentially, you're suggesting that the admin and the executive staff, the C-suite, basically manages this virtual CEO.

11:06David Leary: All right, David, enough about that. I want to talk about these IRS calls. You've got that story. But first, let's thank our next sponsor, and that's Numeral. Let me ask you something. When you recommend a sales tax solution to a client, are you confident in the support they'll get? What about the pricing? What happens if something goes wrong? As accountants, you've seen it all. Clunky support, surprise invoices, and nobody picking up the phone when there's a problem. Numeral is built differently. It's a modern sales tax platform built for accountants, and it supports registrations, filings, nexus tracking, exemption certificates, and even VAT and GST in over 70 countries, all with great support, simple transparent pricing, and no surprise invoices.

11:49David Leary: Then there's the guarantee. If your client's sales tax isn't filed on time, Numeral pays the penalties and interest. And for accounting firms looking for a true partner, Numeral works alongside you to support your clients as they grow. Whether you're implementing it yourself or making a referral, Numeral has you covered. To check them out, go to accountingpodcast.promo.com That's accountingpodcast.promo.com All right.

12:19Blake Oliver: So yeah, let's talk about this, uh, enrolled agent. So this is on the tax girl.com blog. Uh, Angeline, I'm going to, her last name is balstices possibly. I'm not pronouncing that correctly, but it's easy. Tax girl.com. Very easy to get there. Right. Uh, so she's an enrolled agent and she tracked 188 calls to the IRS over four months. And she figured, she figured out that she spent 92 hours on the phone. This is that's two full work weeks. So if this was just four months, that's a third of the year. That means enrolled agents are spending six weeks a year on the phone with the IRS. That's kind of crazy.

12:54Yeah.

12:55David Leary: I mean, it might be seasonal, right? But even if it's just like two weeks out of four months, what, 16, 17 weeks, and you're spending two of those on the phone, waiting around? I imagine a lot of it's waiting around, right?

13:07Blake Oliver: 94 hours on the phone, 74 hours were just waiting, just waiting. So a four to one ratio for every minute of actual conversation, there's four minutes of waiting. Call times really spiked and jumped in September up to 77 minutes versus July's easier times. One in five calls is never connected to anything at all. So 20 % of the times you call, nothing happens. It's a dead end.

13:36David Leary: You just get disconnected? Yeah. I wonder what's happening there. Is that agents at the IRS are in the call center and they're just picking up the phone and hanging it up. Why don't these calls get connected to anyone?

13:49Blake Oliver: Or just ring, ring, ring, and ring? They route it?

13:51David Leary: I don't know.

13:52Blake Oliver: Yeah, it's not really clear. And then, this is really interesting because, did you see they relaunched America.gov? And it's supposed to be, basically, it's a chatbot for the U.S. government to find resources. Now, it's not super useful. It just gives you the links to the IRS. You can ask questions about your tax return. It doesn't know anything. It just sends you to the IRS website. It gives you the links. It's a search engine, basically. It gives you the links so you can go to the IRS website. But it would make sense if agents could ask and do something else while they're asking through a chatbot or an AI that can access the records on the back end.

14:35Blake Oliver: It would make a lot of sense. Like the spirit of the America.gov, like I get it in theory, but it's really just, it doesn't actually help you. It's just a search engine. You can get the same thing in Google.

14:47David Leary: Yeah. If you could chat with the IRS, you could have multiple chats going on for different clients. And you're not sitting around on hold all the time. I mean, this is crazy. The average call lasted 36 minutes, but that was only seven minutes of actual conversation in that 36 minutes. The rest of it's just waiting around on hold. I mean, like, are these agents like taking breaks and just leaving people on hold, just screwing around? Or is it really that hard?

15:19Blake Oliver: Well, sometimes they have to put you on hold to go research. Hold on, I've got to research this, right?

15:23David Leary: Yeah.

15:24Blake Oliver: But it's just, these numbers are, it's just not very productive. I mean, we're wasting the time of lots and lots of enrolled agents.

15:31David Leary: Now, this is important to note. this is not the public IRS lines used by most individual taxpayers. This is the practitioner priority service line. So imagine what the regular lines are like.

15:44Blake Oliver: I haven't called the regular line in a decade at least. Yeah, I can imagine.

15:50David Leary: All right, David, let's go to some more IRS news. I've got some follow-up. Okay. This is about ICE. Remember how the IRS started sharing information with ICE? Yeah, I broke that story.

16:04Blake Oliver: I broke that story months and months ago. Nobody in the media was talking about it.

16:09David Leary: Well, that got appealed. And the U.S. Court of Appeals for the D.C. Circuit just upheld a lower court order barring IRS data sharing with the Department of Homeland Security's Immigration and Customs Enforcement while litigation continues. So that is on pause. This is funny. The IRS, oh, no, this is from the court. The court dismissed the administration's claim that the injunction was burdensome as, quote, weak sauce, unquote.

16:44Blake Oliver: Weak sauce.

16:45David Leary: I want to know who this judge is that wrote this opinion. Yeah, weak sauce.

16:51Blake Oliver: But is it too late? I mean, they asked for 12 million records, I think, originally, and they were handed over 45 ,000. I mean, that's probably going to keep ICE busy for years, to go after all those people if they want to go after them.

17:03David Leary: Although it's so strange. All the news about ICE, I mean, that was a big deal here in Phoenix, given how many immigrants we have, how close we are to the border. And that's all stopped. I'm not hearing about raids, nothing. It's like the administration - I haven't seen them.

17:15Blake Oliver: I think it's because of the midterms. I haven't seen a MAGA hat in like months. Like where did everybody go? Where did all the pro-Trumpers, MAGA, Doge people go? They all vanished. Please put your comments if that's you. But yeah, I think everybody's dialing this stuff back because of the midterms. And then if the people who want to be in the power win the other position, they'll wrap it back up again because there's no threat of losing their seat, right? It'll dial back up, no doubt.

17:43David Leary: I've got some more IRS news. The Senate just unanimously approved a 65 reform IRS modernization bill. I don't think I said that right. It's an IRS modernization bill that has 65 changes in it. And it's called the Taxpayer Assistance and Service Act, bipartisan package intended to modernize the IRS, strengthen taxpayer rights, improve tax administration, and target predatory preparers. The ICPA has praised its preparer oversight, hardship protections, transparency, and service reforms. It now has to go into the House where it could get divvied up and provisions taken out. Here's what's in there.

18:29Blake Oliver: I feel like things like this have passed before.

18:31David Leary: Well, I think we've been hearing about it maybe getting there. This legislation takes a while, I don't know. But here's what the bill mandates. It mandates barcodes for electronically prepared returns printed on paper, and OCR is similar technology for manually prepared returns and correspondence. A public weekly backlog dashboard would appear when items remain unprocessed beyond 21 days or the IRS fails to answer 30 % of toll-free calls in a week. So there's going to be a public dashboard. I like that idea. Transparency, right? We don't have to rely on one enrolled agent

19:07Blake Oliver: reporting her time tracking on IRS calls to figure out if they're doing their job.

19:14David Leary: There's going to be taxpayer portals that show processing milestones, expected refunds, and reasons for suspended returns. And the IRS would offer callbacks when waits exceed 10 minutes. Callbacks, not having to wait on hold, that could cut a lot of that time. Taxpayers could view six years of filings and correspondence, upload replies, and give representatives streamlined access to multiple client accounts. Math error notices would require plain language explanations, itemized adjustments, a helpline, and a bold 14-point abatement deadline on page 1." There's some other stuff in here that I'm going to skip past.

19:53David Leary: The one that is really interesting is the paid preparer crackdown. So this is an ongoing issue. Paid preparers are often dishonest preparers too much of the time. Now, if you want to get a PTIN, you have to get checked. Your own tax compliance gets checked. Plus, you get 18 hours of ethics and tax law education. The IRS could suspend or revoke PTINs and impose penalties up to$1 ,000 or$5 ,000 for fraud, and willful P-10 misuse or evasion would become a felony carrying up to two years in prison and fines up to$50 ,000 for individuals or$100 ,000 for corporations. Misappropriating a client refund would carry a penalty equal to the greater of$1 ,000 or the full refund.

20:44Blake Oliver: And those tax preparers, that's like a certification or a bar that's lower than EA, right? You don't have to do much, is my understanding, to do it, right? To sign off and prepare.

20:56David Leary: Yeah, but well, wasn't there that, there was a case, right, where all that got reversed. The IRS couldn't regulate preparers. So, like, I don't know what the situation is right now, but it's not, it's not, there's not a high bar.

21:10Blake Oliver: It seems like at a minimum, if you're going to do more than one return, you probably should, if you're going to prepare more than one return for somebody else, you probably should get a P-10. I think I could, that would make sense to me. Like you need to register. If you're going to do two returns for two different individuals that are not you or family, yeah, you got to join the party here and register for a P-10.

21:33David Leary: S-corp election deadlines would move to the return due date, including extensions. That's a nice change. Estimated payment dates would also shift. They'd go from June 15th to July 15th and from September 15th to October 15th. Paying 125 % of the prior year's income tax liability would create a failure to pay safe harbor. Partnerships with more than 100 partners would be required to file electronically. The IRS could extend that mandate to partnerships exceeding$100 million in assets or income in the prior three years. I think that's everything worth noting in this bill.

22:11Blake Oliver: So was any money designated to do this modernization and to make these changes? Is there any budget funds available or is this just go do this? They didn't

22:21David Leary: get any extra money, but in a recent bill passed by the House, the IRS got an$11.2 billion annual budget approved through December 11th that keeps it steady. So they've got the same amount of money, at least through the end of the year pretty much, but no extra. All right, I've got more tax stories, David, here. Here is a story about the GAO's, the Government Accountability Office, the GAO's new report estimating the cost of annual federal tax fraud. This is the first time they've done this. And the estimate is in the hundreds of billions, between 116 billion to 304 billion is lost due to fraud, just federal taxes.

23:13David Leary: That is roughly 2 % to 6 % of federal taxes owed.

23:18Blake Oliver: Is this just like they did fraudulent tax returns or is this just in general, the fraud across the economy is causing this loss?

23:27David Leary: It includes return preparer fraud, but also tax evasion and identity theft refund fraud. So we've never had an estimate of this before. We've had estimates of the tax gap, which is the difference between the tax owed and tax paid, which I think has been like$500,$600,$700 billion. This is specifically fraud that covers tax evasion, return preparer fraud, and identity theft refund fraud. They did it with a statistical simulation that combined IRS fraud data, tax gap info and estimates of evasion in the shadow economy. So it's not precise. It's an estimate. And it's not the same as the tax gap, right?

24:06David Leary: That's taxpayer errors are included in the tax gap, fraud or noncompliance, that sort of thing. So fraud is estimated at between 17 and 43 % of the gross tax gap. So let's just say roughly what, like 20 to 40 % if we round. and they estimated the tax gap. Well, the IRS did this for that same year as$606 billion. Oh, and I should say this is for tax year 2022. They tend to lag a few years because they don't have all the data, right? Yeah. So there's that. Also, here's another stat on how much tax complexity costs the US economy. Tax complexity now costs the US economy over half a billion dollars annually.

24:53David Leary: 544 billion dollars. The Tax Foundation put together this analysis. Americans are going to spend 6.9 billion hours on IRS paperwork this year. That's the same as 3.32 million people working full-time all year doing nothing but tax forms. Yes. 3 million people. And that's how you put the dollar value on the time and you get that estimate of lost productivity. It's almost, it's getting close to 2 % of GDP, 1.7%. And it's 28 times what the IRS itself spent in 2025. businesses carry most of the load the average 1040 takes 12 hours the form 1120 for corpse averages 90 hours for large corporations it's 610 hours that's more than 15 full-time weeks on one

25:45Blake Oliver: return these are all numbers that should be on this new website not just the call hold times all these numbers should be on there all right david i'm just going to keep going through my

25:55David Leary: IRS stories. Okay. And then I'm going to hand it back to you. I got to get this done. Okay. So these are two kind of wild stories. One is about a victim and one is about a fraudster or let's say a tax evader, somebody who is a little too aggressive. And I want to thank Ray Arellano for sending this to me. This is kind of a, it's relevant, David, because it's about an influencer who went to tax court. And tax court disallowed this interview's influencers celebrity access deductions. The case is called Sammy versus Commissioner. Suleiman Sammy worked full-time in IT while operating three side hustles, transportation, event ticket sales, and social media influencing.

26:44David Leary: And his influencer content featured athletes and celebrities. He sometimes appeared on camera, sometimes behind it. This was during 2019 to 2021, and Sammy paid thousands of dollars to attend the Grammys, the Emmys, Tiger Woods Tiger Jam, other events offering celebrity access. He paid to meet Benedict Cumberbatch, Matt Damon, and Mark Ruffalo, and he bought a personalized video from Chris Evans. He also paid to catch passes from Tom Brady and Drew Brees. He fumbled Brady's pass, but he did catch Breeze. And the argument for these being a business expense was that these celebrity posts attract more views and attention.

27:29And Sammy argued in court that spending this money, taking this deduction, was to get shared

27:36David Leary: advertising revenue from his growing social media following. He called them marketing expenses, but the IRS challenged them. and the tax court disallowed the deductions, finding that his primary motives appeared personal or charitable rather than sufficiently tied to business marketing. And he generated, I think this is kind of the key sticking point. He generated no influence or income with those expenses. So he's spending that money to build up the social following, right?

28:09Blake Oliver: Yeah, this is why it's not necessarily in ordinary because the business didn't gain from this behavior.

28:16David Leary: Well, here's the thing is he did start making money in later years because it took him time to build up this following. But the court didn't, I guess, think that was important, irrelevant or considerate.

28:29Blake Oliver: Well, shouldn't he have like capitalized that or something and then like make it a capital expenditure, then expense it when he actually gets the revenue from these things he's paying for and then maybe it would have been okay. if it was aligned with the revenue?

28:43David Leary: Yeah, maybe. So I guess there was some other thing in here that was funny is that he's actually an accountant. Oh, of course. Where did he work? He worked at, like he should know better, right? I can't find it right now in the story. But yeah, he has two accounting degrees and he didn't keep good records.

29:15Blake Oliver: I feel like this is like, make an example of him. He should know better. This is ridiculous.

29:21David Leary: No. So apparently measurable audience growth does not necessarily establish the business purpose, especially when the expenses are really, really fun related. Arguably frivolous, yeah. One more story here. A North Carolina woman, she lost$600 ,000 in a romance scam. I saw this on CPA Practice Advisor. The IRS then sent her a bill for$225 ,000 in taxes and penalties. I'm rounding there. Why did that happen? Because the money she sent the scammer came out of her retirement account, so it counted as taxable income. Her name is Lori Flowers. She spent months talking online with a man who said he was from Belgium and was going to move to be near her.

30:12David Leary: He convinced her to pull money from her retirement savings and sent it to him. We call that pig butchering. He built trust for months, then he took her money. And part of the problem is that, and why she owed this big tax bill is back with the 2017 tax law, which got rid of the theft loss deduction that fraud victims could rely on. So if you lose money that you pulled out of your retirement account, you still owe tax on it. There's an act in the House that passed the House that's sitting in the Senate that would fix that. But at the moment, if you have parents that fall for some scam like this and do that and take a penalty, you're out.

30:59Blake Oliver: I feel bad for her, but she knows those withdrawals are taxable, right?

31:04David Leary: Well, not everybody. Not everybody knows this, you know.

31:08Blake Oliver: Let's say she didn't get pig butchered and he actually moved there and they're truly in love and they're happy they're ever after, but she was still making the withdrawals. Like it's kind of on her a little bit for not knowing, you know, or her financial advisor, whoever set up this retirement account for her.

31:24David Leary: All right, David, I'm going to pass it back to you after I thank our next sponsor, and that is Savant Labs. If your finance team is using AI for drafts and research but still closing the books by hand, you're in the same spot as most enterprise accounting teams. Claude and Copilot handle the easy stuff. But what about the close, the reconciliations, and the tax provisions? Still manual, still spreadsheets, still eating up days every cycle. The real issue is an effort, it's trust. General purpose AI gives you a different answer every time, and an auditor can't follow the trail. That's why Savant exists.

32:02David Leary: You describe a finance task in plain language, and Savant turns it into a governed agent. It pulls, cleans, and reconciles your data across ERPs, spreadsheets, and PDFs with over 500 connectors. Every workflow runs the same way every time. Same inputs, same output, every cycle. Data lineage approvals and SOX reports are built in automatically. The result is up to 70 % less manual work and 80 % fewer errors. Here's the best part. Bring Savant a complex use case that's eating 20 plus hours a month, and they'll build you a custom agent for free with a 30-day trial. To learn more about Savant and start building your first agent, go to accountingpodcast.promo.com.

32:45David Leary: That's accountingpodcast.promo.com. I have an idea for your first agent, one that calls the IRS and

32:53Blake Oliver: sits on hold for you. It could be your first agent. It's just constantly -

32:57David Leary: Well, there's apps for that. I think there's services for that, yeah. Yeah. It exists.

33:01Blake Oliver: All right. So do you remember two or so months ago, I had a story about the two junior EY employees that they allegedly accessed during the audit, the bank account and personal bank account of the Australian prime minister, Anthony Albanese. Do you remember this story?

33:20David Leary: They're auditors and they had access to like, They were auditing a bank, right? Something. And they grabbed his statements or something.

33:27Blake Oliver: Yeah. So they got caught. They were fired. Charges are being issued against the two men, age 21 and 25, right? Because they've accessed this bank data they weren't supposed to access. Okay. Fine. Well, last week, we didn't get to this story. And now I'm glad I didn't because it ties into a bunch of stories this week. Apparently, OpenAI's agent breached the Australian government's system for Medicare data.

33:54Blake Oliver: So no charges, there's no arrest, right? So two junior accountants get fired for peeking at a bank statement. Open AI's agents hacks into the Australian Medicare system and no charges. So that was a story I had last week. So I was kind of watching that.

34:12David Leary: This is happening because they're testing these frontier models on finding security gaps because that's what they're really useful for is plugging all these holes that were built or that didn't get fixed when they were built by humans. And these models are amazing. They can run penetration tests and solve it and fix it. So they have to test. They're testing it on test companies, but then these models will break out of the test environment and go hack real servers and stuff.

34:42Blake Oliver: And it's worse than that. Apparently, OpenAI's agents have now accessed apparently more than 100 organizations they've had to notify about things. I'm willing to bet it's thousands at this point. And I also want to bet before the year is over, we are going to cover a story how OpenAI's agents access the IRS. I 100 % believe this. We should just get the episode ready to go now and put a date on it. We will be talking about that story. There is zero doubt. So basically, OpenAI sent out alerts to more than 100 organizations after discovering the agents. So basically, after the hugging face attack in July, they've been doing all this research internally to figure out what happened when, et cetera.

35:24Blake Oliver: They have 50 petabytes of data they have to go through to paint a picture of what a petabyte is. It's a basically 36.4 billion floppy diskettes, you know, the three and a half inch floppies, it is an amazing amount of data. Now they, they say they're going to go through this and they're, I think another story I saw, they're going to put 7 ,000 servers on this processors to look at all this data to figure out what happened. But I'm kind of starting to form the opinion that Hugging Face is the only company that had the skill set to detect it. Because remember, Hugging Face detected it and then they had their agents fight it back kind of a thing, right?

36:05Blake Oliver: I think all Fortune 500 companies, I think they've all been accessed now. I think 80, 90 % of them, they have not detected it. There's also an article that came out. So US government things. So it accessed the US Department of Education, the Commerce Department, the SEC, all without anybody's knowledge. So the IRS is right around the corner. We are going to find out this story.

36:30David Leary: Well, no, the thing that protects the IRS is the fact that their systems are so old, you can't log into them. You have to be in that office and walk over with a disk and stick it into the computer. Possibly. They're not even networked.

36:45Blake Oliver: Who knows? to believe this we'd like to believe this right but like in the way if you step back and think about this because they remember you were talking about like two weeks ago about it's not fair like somebody does this bad fraud they get fine they don't get fined as much maybe you're talking about that like the punishment doesn't like fit the crime and i think about this a lot like that's kind of what's happening here right these these two ey kids peek at a bank account bank statement and they get fired and have charges brought against them. There's no charges happening here because all these frontier models are saying it's a lab.

37:20Blake Oliver: We're testing in our lab. But think about this. Ford. Ford crash tests their cars, right? What if Ford's cars were just driving on regular roads running into us? That's what's happening right now. It's not acceptable.

37:35David Leary: The analogy is not quite right because as far as we know, these models haven't leaked data. Oh, no, that's not true.

37:42Blake Oliver: They have leaked data. They have. Yes, they have. They've captured photos and released photos. Where did it go?

37:51David Leary: Like taken, hacked into systems and then released information online that other people were able to get? Because that's the crime, right? Yeah.

37:59Blake Oliver: They somehow posted 53 user-provided images, like maybe images you've uploaded to OpenAI and then they reposted them on other sites. It's not like...

38:09David Leary: but not something big like you know like leaking taxpayer information out of the irs like uh yet

38:16Blake Oliver: yeah yeah i think we're in the yet on this because they think about all basically they have all this data they have to go through now yeah it's a it's a crime scene investigation essentially but there's no charges like i guess he's going to go through the data they're going to just have ai agents go through the data well that that's what's funny too they uh i had a note on that they they're claiming that it's going to take months to investigate. And I'm like, wait a second, don't you have the greatest, smartest AI being that's ever existed? How do you not do this quicker? They're going to dedicate 7 ,000 high-end GPUs to go through this 50 petabytes of historical data.

38:55Blake Oliver: We're going to find out thousands of companies have been accessed. Hugging Face just happened to be the only company with the skillset to detect it and stop it. Nobody else is. And, and yes, they're close. They might be helping companies discover other holes and security patches might exist, but that's just a, that's just there. Like, of course, if you get caught hacking into somebody, you're going to say, I was doing it to help them, let them know that they had a hole in their fence or their block was unlocked. You're going to lie right there. I, this is like, it's frustrating. It's really crazy that there's this much happening, but mark my words, some barker words I've said here.

39:32Blake Oliver: We will be covering a story about open AI access in the IRS before the year's over. Zero down. I guarantee you it's already happened. How can I not? You're like a nerd tech guy in this lab. Oh, what could we, what could it hack the IRS? Let's see. It's, you know, they type that in as a scenario and of course it went and did it.

39:51David Leary: All right, David, it's time for you to put your money where your mouth is and start a prediction on one of those prediction markets.

39:56Blake Oliver: Prediction market, yeah. I'll vibe code my own prediction market on OpenAI's hacks.

40:02David Leary: Checking in on the live stream here, we've got Tyler. Tyler Otto says, to be a tax preparer, it's like$85 a year in P-10 renewal fees, be a US citizen and not have a felony. That's the bar to being a tax preparer. Thank you for updating us on that. He says, it's a pretty low bar, but I guess it would keep my uncle out. So that is saving America.

40:26David Leary: Stanley says, they are telling the AI agents to hack. So he's with you that this is not just an accident, that they're using the cover of being a lab in order to poke around, mess around. Fez says, Tyler, I know him. He's like famous. All right, David, how about you thank our next sponsor while I tee up our next story? And that sponsor is going to be Cloud Account and Staffing. make sure you take a big swig of water before you get into it because there's nothing worse than getting halfway through an ad read and then your throat gets dry and you get nervous. That happens to me.

41:09Blake Oliver: Are you tired of the endless search for qualified accounting talent? You're not alone. Growing accounting firms are struggling to find available and affordable team members when they need them most. Clyde Counting Staffing has the solution with the revolutionary candidate portal. Unlike traditional staffing agencies that waste your time with sales calls, paperwork, and deposits, the Cloud Accounting Staffing Candidate Portal gives you instant access to highly vetted, qualified accounting professionals. No waiting, no hassle, just top talent right now. What makes this different? Speed and simplicity.

41:40While other firms make you wait weeks or months with Cloud Accounting Staffing, you could interview somebody as soon as tomorrow.

41:46Blake Oliver: Their boutique support ensures you're getting quality talent that's both available and affordable, exactly what a growing firm needs. The Candidate Portal puts you in control. Browse live candidates, make selections on your timeline, and build your offshore team without the traditional headaches. To find review and book interviews with potential team members, all in less than 10 minutes, head over to accountingpodcast.promo.cas. That is accountingpodcast.promo.cas.

42:13David Leary: I want to talk about big law. and the billable hour because they're having some challenges that might be informative for big accounting and just accounting in general. This came from DealBook in the New York Times. My dad sent it my way. So big law has really embraced AI. By big law, I mean these giant law firms that employ thousands of lawyers. Sullivan and Cromwell now has an agreement analyzer that reviews deals. Cooley has something called GoPublic that assists with S1 drafting. Ropes in Gray is developing a system that can produce detailed diligence reports in hours instead of weeks. There's also a bunch of legal AI products from Google and Anthropic.

42:59David Leary: Lots of investment here. And the report from the field is that clients are increasingly asking what this AI efficiency means for their bills because those legal bills from those big law firms are eye-wateringly high from an hourly standpoint. Over$1 ,000 an hour on a lot of these attorneys. And some of these clients are now using AI agents to identify invoice entries where automation could have reduced hours and then are striking those charges or demanding discounts. Kyle Poe of Legora said that in-house legal teams often request 20 to 30 percent across the board reductions now. And the firms are saying, well, AI hasn't actually saved us that much time.

43:47David Leary: And, you know, it's letting us do more substantive analysis, but they're still having to fight this pressure on fees. And they are facing a bunch of AI native competition. AI native meaning these legal providers that are using AI a ton from day one, they are using upfront pricing, not hourly billing. So Norm AI, they have an affiliated law firm and they serve financial services clients. So it's an app plus lawyers. They raised$260 million in July at a$1.2 billion valuation. And they're going after big law. And it puts these traditional firms in a bind. If they use AI, it reduces their billable hours, which reduces their income, but not using it allows competitors to undercut them.

44:39David Leary: So what are they going to do? They're going to have to shift their pricing. Citi did a survey of 55 large law firms, and 35 % of those expect to modify their billable hour models within the next year because of AI. That's over a third of these big law firms are going to switch to something other than billable hours, it seems, or do something in that regard. And it goes to two thirds of them by 2035. Wells Fargo, they said that large firm revenue grew more than 12 % on average during the first six months of 2026, while standard billing rates rose by double digits. So fees have still been going up, right?

45:21David Leary: But I think that we're going to see that really get disrupted.

45:29Blake Oliver: I feel like it was a tweet or somewhere that, or maybe we talked about on the show, I can't seem to find it, but that law firms are realizing the hours they can bill for is going to decrease. So they're just changing the rate. Like, well, I'm just going to make my rate a$4 ,000 rate instead of$400 an hour now. Yeah.

45:49David Leary: But that breaks at a certain point because clients see the higher rate, they see the increases on the rates. And that is what makes them go check out these new startup law firms that offer them fixed fees upfront, or something different than an hourly system. So I think it is going to break. And law firms are going to totally pivot, like we saw two thirds of them by 2035. So, you know, like that's within the next decade, they are going to switch away from their hourly billing model. And I think accounting will follow. So Ron Baker -

46:29Blake Oliver: It's either switch your billing model or - Stay with us, Ron Baker.

46:33David Leary: Stay with us, Ron Baker. You will see it happen during your lifetime. Stick around. Don't give up on us.

46:39Blake Oliver: or you have to add more services. Like for example, Soarin, which is a top 50 accounting firm. We talked about them about a year ago. They rolled up a bunch of regional firms. Yeah, they're a private equity roll up. Yep. Roll up, right. They now have launched Soarin Law in partnership with Mansfield and Mayo. So just like Aprio did this similar move earlier this year. It was actually Aprio in Arizona, right? That I think did that first.

47:02David Leary: Yeah, because Arizona now allows you to do that.

47:05Blake Oliver: Yeah, because that's the only way you'd be able to weigh. Like if you're like worried about your revenue going down, you're going to have to sell twice as much stuff to what the same customer and sell them legal and accounting services. This makes a lot of sense. And honestly, like from the other side, like I just want to go to one place and have it all done. I don't want to have to go to separate entities to get my legal work done. And it gets rid of that dumb dance. All these accounts are like, Oh, I'm never going to help you incorporate. That is not accounting job. That's legal work. Like there's like, just get rid of that divide.

47:34Blake Oliver: Cause at the end of the day, the business owners don't care. like this constraint, like that's legal work. We're not allowed to do that. We're accountants. This just gets rid of all that. All right.

47:47David Leary: Illinois ordered an audit of AI developers, but who's going to do it? That's the question in the Chicago Tribune. Here's the story. The state of Illinois just became the first state to require independent audits of AI companies' safety claims. starting in 2028 the big ai developers are going to have to bring in an outside party every year to check whether they actually followed their own safety plans and sign their name to the answer

48:16Blake Oliver: what do they mean when they say big ai developers they're talking about like the frontier models

48:21David Leary: open ai anthropic yes open ai and anthropic they actually back the legislation these big companies

48:30Blake Oliver: Oh, interesting.

48:31David Leary: And you might wonder why would they do that? I think it's because once you get big enough, more regulation stifles competition.

48:38Blake Oliver: I fully agree on that, yeah. And it also helps them put aside these bad news stories about safety, like what we just

48:47David Leary: talked about, these AI models leaking out of the lab and hacking into real companies. So this is going to be an annual independent audit. It's going to examine whether they followed their published safety plans. An outside party will sign on the conclusion. And California and New York are doing similar things. They're making developers publish safety frameworks and report serious incidents, but there's no audit required as part of those. Illinois is the first state to require an audit.

49:19Blake Oliver: Well, instead of doing an audit, why don't they just look at their internal systems and discover that they've already accessed them? What are you going to audit? Just look at your own systems. Guaranteed the state of California for sure has been accessed, no doubt.

49:33David Leary: The auditors have to follow, under this law, generally accepted auditing standards. And here's the crazy part, though. Like, they don't actually have to be auditors like we think of them. You don't have to be a CPA under this law to do these audits. There's no license, certification, registry, accrediting board, or disciplinary body for these AI auditors. And this would be perfect work for CPAs. They just didn't require it. So –

50:09Blake Oliver: Because it sounds a little bit more like an internal controller type or internal audit type thing, even though they're not –

50:17David Leary: Well, they're external though. These are external third parties during this audit. And they use generally accepted auditing standards as the standard for this audit. But they didn't actually require that anyone hold a license. So basically, it seems like anybody can do it. Tyler's uncle can do it. Yeah. So I don't know. Maybe I'll just like – and here's the thing. Think about this. OpenAI and Anthropic, right? They're going to go and hire a company to do this, a company or an auditor that doesn't have to have a license. Who do you think they're going to hire?

50:49Blake Oliver: This is going to be the crypto companies all over again when they got the fake attestation reports.

50:56David Leary: Attestation reports, right? The non-audits that they called audits? Yeah. Yep. So, I mean, this is a big missed opportunity here. I mean, without requiring that these people have licenses or professional competence, like something that ensures that they're actually independent third parties, like this is just a giant smoke and mirror situation. Yeah. So the public's going to be getting these reports and feeling better about AI security or whatever. But we all know that when anyone can sign an audit, it doesn't mean anything necessarily. Like the whole point is trust. So if you don't like limit the pool of these potential auditors and ensure that they're professionals and they're ethical and they're competent to some extent.

51:44David Leary: You get the same situation as the paid tax return preparer situation.

51:50Blake Oliver: Yeah, this is probably a situation where the AICPA had a chance to make a letter in response to this story, issue an opinion on this, right? Yeah. For the profession.

52:01David Leary: All right. Let's see. David, I have to jump. I got to go.

52:06Blake Oliver: I'm going to do it in five seconds. Remember I talked about Meta? Facebook launched Meta? Yes. I'm sorry, Meta launched Muse. uses the new AI and I said like it's the microwave moment for this. It didn't take long. Four days later, they actually have direct integrations to QuickBooks, Shopify, Stripe, Slack, and a dozen other back small business owners apps. So they're going to take this from being your personal assistant and now it's your small business assistant overnight, just like that.

52:29David Leary: All right. Thanks everyone who joined us live. Great to have you here. Thanks for chatting. Subscribe on YouTube, hit that notification bell icon. You can get notified when we go live and join us. You just got to find the Accounting Podcast on YouTube. You can find out what we look like. Sometimes it's really surprising. People are like, what? I've been listening to you for all those years. I get called Blake.

52:50Blake Oliver: People finally see you in person. They call you the wrong name. Yeah.

52:55David Leary: And we'll see you here next week. I am, where are we going? Later this month, we will be at Intuit Connect. I will also be at CCH Connections in Orlando. So it's going to be a busy week for me, end of October. If you will be at any of those events, let me know. We could meet up. I'll be at CCH Connections on Monday, the 26th, and Intuit Connect on Tuesday night and Wednesday, the Tuesday, the 27th and Wednesday, the 28th. You'll be there, David. And so send us an email, the accounting podcast at your mark.me and hunt us down. Thanks, David. Thanks everyone who joined us. Thanks to all of our loyal listeners for joining in on episode 508.

53:45David Leary: And we'll see you around here next week.

53:55Blake Oliver: Time for the classifieds.

54:01Blake Oliver: Can you defend your client's Shopify margin number? ProfitRoot turns a Shopify orders export into a reconciled margin report. Every figure traces back to the source file, every coverage gap is named, and nothing is estimated to fill a hole. There's no sign up to try it. Walk the preloaded sample store or upload an export and the first report is free. Built for accountants and fractional CFOs, Try it at tryprofitroot.com.

54:53Thank you.