I’ve been getting more of these lately. A Slack message or email arrives, I open it, and there it is — several paragraphs of clearly AI-generated text, heavy formatting, bullet points, em-dashes. The person who sent it probably spent about ten seconds on it. They asked a chatbot, copied the output, and hit send.
That has a name now: sloppypasta. Verbatim LLM output copy-pasted at someone, unread, unrefined, and unrequested. And it’s becoming a real professional etiquette problem — especially for CPAs and finance professionals who work in trust-based environments.
What Actually Changed
For most of human history, effort-to-write balanced effort-to-read. If someone sent you a memo or a report, you knew a human had invested real time in it. The writing itself was a credibility signal.
LLMs broke that signal. You can generate a five-page essay in seconds. The writing cost is now zero for the sender — and transferred entirely to the recipient, who still has to read it, vet it, and figure out if any of it is actually correct.
That’s the core problem. It’s not that AI is bad. It’s that forwarding AI output you haven’t read is asking someone else to do the work you chose not to do.
The Three Archetypes
There are three ways this shows up in professional settings:
The Eager Beaver
Well-meaning. Someone jumps into a team discussion, asks a chatbot, and dumps the generic response inline. The intention is good — they want to help. But the wall of unrelated AI text blocks the live conversation. Everyone else has to scroll past it or stop to vet it. It feels helpful to send. It creates work to receive.
The OrAIcle
Someone asks a specific question. Instead of answering, this person pastes “ChatGPT says…” with a generic response that ignores the actual context. It’s the LLM-era equivalent of LMGTFY — at least that was honest about what it was. Sloppypasta looks like a real answer. Recipients have to do extra work to figure out it isn’t one.
The Ghostwriter
The highest-risk archetype for CPAs. This is someone sharing AI output as their own work, with no disclosure. Recipients have no reason to question it. They may act on information that is out of date, wrong, or hallucinated. And when it fails, it’s the sender’s credibility that takes the hit. Using AI as a ghostwriter borrows your credibility. If the content is wrong, that’s what gets spent.
Why It Matters More for Accountants
Two specific risks that CPAs need to think about:
Professional liability. In accounting, your name on a document implies you stand behind it. AI ghostwriting without disclosure violates that implicit contract. A hallucinated figure in a tax summary, an outdated regulatory reference in a client memo, a wrong jurisdiction in a financial analysis — these aren’t just embarrassing. They can carry real professional consequences.
Audit trail and documentation standards. Accounting engagements require documentation of sources and methodology. AI-generated content with no disclosure, no source attribution, and no verification process doesn’t meet that standard. Firms building AI use policies need to address not just what AI can be used for, but how AI-sourced content must be documented when it informs client work.
Six Rules for Using AI Responsibly
- Read it. If you haven’t read the output, you don’t know whether it’s correct, relevant, or current. Don’t forward what you haven’t read.
- Verify it. Anything you share carries your implicit endorsement. LLMs produce outdated facts, wrong figures, and plausible nonsense. Check the numbers. Check the dates. Check the jurisdiction.
- Distill it. Cut it down to what matters. Sending four paragraphs when one would do isn’t helping anyone. This is also your chance to put it in your own words — which is how you demonstrate you actually understand it.
- Disclose it. If you’ve read, verified, and edited it, send it as yours — with a note that you used AI assistance. If you’re sharing raw output, say so. Disclosure restores the trust that sloppypasta destroys.
- Share only when requested. Don’t drop AI output into conversations unsolicited. The Eager Beaver pattern is well-intentioned but creates real friction. If someone didn’t ask for a research report, don’t give them one.
- Share as a link. If it’s a longer piece of AI-assisted research, attach it as a document or share it as a link — don’t paste fifteen paragraphs into a chat thread. A link lets the recipient choose when and whether to engage.
The Bottom Line
AI capabilities keep increasing. Using these tools to draft, research, and accelerate your work is genuinely valuable. But using AI should not make your productivity someone else’s burden.
New tools require new manners. Use AI to improve what you send. Don’t use it to replace thinking about what you’re sending.
Key Takeaways
- Sloppypasta is verbatim AI output forwarded unread — it transfers the verification burden to the recipient
- LLMs are economically incentivized to produce longer, more formatted responses — volume does not equal accuracy
- The Ghostwriter archetype carries the highest professional risk for CPAs: borrowed credibility, no disclosure
- AI-sourced content used in accounting engagements must meet documentation and disclosure standards
- The six rules: Read it. Verify it. Distill it. Disclose it. Share only when requested. Share as a link.
Want the CPE credit? Take the full lesson on EverydayCPE and earn 0.2 CPE credits: [lesson link]
Leave a Reply