Back to Editorials
Technology 8 min read

I Don't Email That. Neither Should You.

Josh Mauer, CPA
Founder, Josh Mauer CPA LLC

The subject line was the breach

Every filing season, and most months in between, somebody's tax documents land in my email inbox. Not in the portal. In my inbox, as attachments, sent from a personal email account. It is one of the most common things that happens to me, and the people doing it are not careless — they are being responsive. They are getting me what I asked for, the fastest way they know how.

Here is the pattern I want to describe, because the detail is the whole point. Somebody attaches a year-end brokerage tax reporting statement. Whatever they clicked to send it copies the document's header straight into the subject line of the email. So the subject line itself — the part that sits unencrypted in a mail server log, in a push notification on a lock screen, in every forward and every backup, in the index of every mail client that ever touched it — ends up carrying a full name, a brokerage account number, and the last four digits of a Social Security number.

That is not an edge case. That is a Tuesday. And it is worth sitting with what it means: the exact information a criminal is trying to get is not inside the encrypted attachment they would have to open. It is sitting in plain text on the outside of the envelope, duplicated across a dozen systems, readable by anything that can see a subject line.

Nobody opens the attachment to check what got copied into the subject. The attachment is the thing you are thinking about. The subject line is the thing that actually travels.

I am not writing this to embarrass anybody. I am writing it because the people this happens to are sophisticated — business owners, physicians, people who have managed real money for decades — and they are doing a completely normal thing. That is what makes it worth saying out loud. The failure here is not a lapse in anyone's judgment. It is a default.

I don't read the cases. I live them.

There is a version of this article written by someone who read a statistic. This is not that article.

In twenty years of practice, not one year has gone by without walking at least one client through identity fraud. Not a scare, not a suspicious email — the real thing. Returns filed in their name before they filed. Refunds routed somewhere else. Accounts opened they never applied for. Months of correspondence with the IRS to prove they are who they say they are, which is a genuinely strange experience for a person who has never had their word doubted in their life.

Some clients came to us because of it. The fraud was the introduction. That is a hard way to meet your accountant.

Business owners know this better than most, because they have more surface. More accounts, more vendors, more employees with access, more people who can send a convincing invoice. If you own a business, you have probably already had a near miss — a wire request that looked right, an email from a vendor whose domain was off by one character.

So when I decline to email a return, it is not policy theater and it is not me being difficult. It is the one decision in this whole domain that is entirely within my control, and I have watched enough of the alternative to make it without hesitation.

What AI changed

For most of those twenty years, the economics of fraud protected you. Stealing from you required a person to spend time on you specifically, and you were not worth the hours.

That math is gone.

I want to be careful here, because the loudest version of this story is not the accurate one. The common claim is that AI now clones your voice and talks its way past your bank's authentication. I went looking for the evidence, and it is thinner than the headlines suggest. FinCEN's November 2024 alert to financial institutions on deepfake fraud does not list a single voice-specific red flag, and puts no number on it — only that suspicious activity reports describing suspected deepfake use rose from 2023 into 2024.

The documented damage runs the other direction, and it is worse for you. AI is not mainly being used to impersonate you to your bank. It is being used to impersonate the people you trust to you, and to impersonate you to the people who trust you.

The FBI's 2025 Internet Crime Report is the figure I would put weight on: 1,008,597 complaints and $20.9 billion in reported losses in a single year. Within that, more than 22,000 complaints specifically referenced AI, carrying $893 million in adjusted losses. Business email compromise — the fake-invoice, altered-wire-instruction category that lands on business owners — accounted for 24,768 complaints and $3.05 billion on its own. The report is explicit that voice cloning is used to request wire payments in those schemes, and to mimic a loved one in distress in grandparent scams, which victims reported more than $5 million in losses to.

Those are reported losses. They are the floor, not the number. Most of this never gets reported at all.

Which brings it back to your inbox, because that is where the impersonation gets its material.

If something gets into your email, it does not skim. It reads everything — every message, every thread, every attachment — and it does not get bored or tired. Then it knows who your banker is, how you sign off to your spouse, which vendor you pay on the first of the month, what your accountant asks you for in March, and what your kids are named. The next message it sends will be the most convincing one you have ever received, because it will have been written by something that read your entire correspondence first.

The account numbers sitting in your subject lines are just the part it does not even have to work for.

Here is the question I would like someone to answer for me. What is Google's security rating? What is Apple's? How many intrusion attempts do they absorb in a day? How many fail? How many succeed? How many of the successful ones get published?

You cannot find out. Not because anyone is lying, but because nobody publishes it and no standard exists to publish it against. These are excellent companies with serious security teams, and I use their products, and I still cannot tell you the answer. I am asked to accept, on faith, that the place where my clients' documents sit is safe.

I do not build a professional practice on faith. I build it on controls I can inspect.

Fewer places, and a record of who looked

The portal is not about encryption, or not mainly. Encryption is table stakes. The portal does two things that email structurally cannot.

It reduces the number of places your information lives. An emailed document does not go to one place. It goes to your sent folder, my inbox, both mail providers' servers, every backup either of us runs, your phone, my phone, the laptop you read it on at the airport, and anywhere either of us forwarded it. One document, a dozen resting places, each with its own door. In a portal the document has one home, and access is granted rather than copied. The question stops being "who might have a copy" and becomes "who has permission."

It creates a record. Every open, every download, every upload, timestamped. That record does not just protect you from outsiders. It holds the three of us accountable to each other — my team, me, and you. If a document was delivered, we can prove it. If a notice was shared, we can prove when. If something went wrong, we can see where. I would rather operate under that scrutiny than have you take my word for it, and you should want the same from any professional who holds your financial life.

Email gives you neither. It gives you a copy, infinitely duplicated, with no record of who read it.

And this is about to matter more, not less. The portals are getting assistants. There is one on this site right now — go talk to it. In the next generation of tax and finance hubs, you will have one too: something that can actually answer what your position looks like, what is due, what changed. That is genuinely useful, and it is also a system sitting on top of everything about you. Which is exactly what a criminal wants access to. The more capable these hubs become, the more the question of where your information lives stops being paperwork and starts being the whole ballgame.

One portal will not rule them all

Here is where I have to be honest about the limits of my own pitch.

Everybody wants one place. One login, everything linked and synced, the whole financial picture in a single view. Every company building in this space is promising you some version of it, and I understand the appeal, because I want it too.

It is not going to happen. Not from one company, not from me. Your bank will not hand its data to your brokerage. Your brokerage will not hand its data to your payroll provider. Your insurer will not hand anything to anyone. Each of them has its own reasons — competitive, regulatory, legal — and none of those reasons are going away because it would be convenient for you. Anyone telling you they have built the one ring is selling you something.

So here is what I will actually commit to. Now that building software is within reach of a practice this size, I will give my clients as much of a financial hub as I can honestly design — as much of your picture in one place as I can assemble without creating a new vulnerability to get it. That caveat is the job. I am required to put the safety of your information ahead of the elegance of my product, and I would do it even if I weren't, because I have sat with the people on the other side of that tradeoff.

Which means some things will stay separate that you wish were connected. When I tell you I can't link something, that is usually not a technical limit. It is a decision.

The policy nobody tells you to buy

Two practical things, and then I'll stop.

If you own a business, get a standalone cyber policy — and read one specific thing in it. General liability does not cover this, and the cyber endorsement bolted onto a business owner's policy usually covers less than the brochure implies. But the detail that actually decides whether you get paid is the difference between two coverages that sound identical.

Funds transfer fraud is when someone breaks into your account and moves your money. Social engineering fraud is when someone deceives you and you move it yourself. Many policies cover the first at the full limit and the second at a much smaller sublimit — when they cover it at all. The second is the one that actually happens to business owners.

So ask your broker three questions in writing. Is social engineering fraud covered. What is its sublimit against the policy limit. And what verification procedure am I required to follow for a claim to be honored. That last one catches people: some policies require a documented call-back to a previously known number before a wire, and will deny the claim if you skipped it.

And consider one personally. This is the part people find strange coming from me, because I will say plainly that insurance is a house game. It is not a scam and it is not a pyramid — the risk pooling is real, which is exactly why it works. But the house writes the contract, reads the actuarial tables, sets the price and defines the exclusions, and it does all of that for a living while you do it once. Structurally, you are expected to lose. That is not an accusation. That is the business model, and anyone telling you otherwise is selling.

Which is why I buy it in exactly one situation: the loss I could not absorb. I have sat with too many people whose brokerage account was drained, and watched how long it takes to get that money back — and sometimes whether. That is the loss that ends a retirement, not the one that dents a quarter. So I will take a deliberately bad bet on the catastrophic tail and decline it everywhere else. Two sides of the same coin, and I would rather name both than pretend the coin has one.

What I would actually do, in order: turn on multi-factor authentication at every financial institution you hold an account with — not by text message, which is the weakest version. Set up a verbal password with your brokerage so a voice on the phone is not sufficient. Freeze your credit; it is free and it is the single highest-return hour you will spend this year. And stop emailing documents with numbers in them, to me or to anyone.

Patterns are what make it plannable

I will admit I chuckle a little at how this moment gets described.

The Boomers are nearly out. Gen X is either heading for the exit or white-knuckling the next ten years hoping they can keep up. The twenty-somethings are genuinely frightened that the life their parents built is no longer attainable. And in the middle, families are working themselves ragged to pay the bills and raise their kids.

All of which is real. And all of which is also exactly what every generation before this one went through, on a cycle you can trace back as far as you care to look. The technology was different. The fear was identical.

That is not a dismissal. It is the opposite. The reason I bring it up is that once you can see the pattern, it stops being frightening and starts being something you can plan around. Fear makes people do the two worst things: freeze, or move all at once at the top. Pattern recognition lets you do the boring, correct thing on a schedule.

AI is going to unlock an enormous amount for all of us. I believe that, and I am building on it. But like work-life balance, nobody is going to hand you the right configuration — each of us has to find the path that actually works for our own situation, and that takes judgment, not enthusiasm.

So be careful out there. Where you put your information, and what you give access to it, is going to matter more in the next ten years than it has in the last twenty. Start with the easy one: stop sending it by email. Use the portal. It is not bureaucracy. It is the one part of this I can actually control for you.

Sources

  1. FinCEN Alert FIN-2024-Alert004, "Fraud Schemes Involving Deepfake Media Targeting Financial Institutions," issued November 13, 2024. fincen.gov/sites/default/files/shared/FinCEN-Alert-DeepFakes-Alert508FINAL.pdf
  2. FBI Internet Crime Complaint Center, 2025 IC3 Annual Report. ic3.gov/AnnualReport/Reports/2025_IC3Report.pdf