Free Resource
In four weeks, AI disclosure went from ethics debate to infrastructure: Substack scans it, LinkedIn flags it, the EU legislates it, and Anthropic watermarks every word Claude writes. This kit takes your firm from "no position" to "written position" in one sitting.
The Problem
Anthropic's watermark records who typed the words, not who controlled them. Hand-paraphrased junk passes the scanners clean, while the most heavily governed AI content in your feed carries the strongest AI flag. Your firm almost certainly faces no legal risk. It does face a story problem.
The fix is not retyping your work to hide the tool. It is a written disclosure position: who produced the words, who stood behind them, and whose name is on the file. The EU has already defined the safe harbor as human review plus named responsibility. This kit gets that position written before the first client runs a scan.
What's Inside
Eight sections, one sitting, and a statement with your name on it at the end.
Role, published, public interest, used in the Union, and the editorial-control carve-out. Run every output through the gates and see what the EU AI Act can actually reach. Client deliverables exit at gate two.
Who produced the words, and who stood behind them before they moved. Common firm outputs pre-placed: auto emails, reviewed summaries, auto-published social, governed long-form, and automated deliverables.
Three sentences, adaptable, usable verbatim in Substack's "How I make this" field, your website, and your engagement letter. Plus the variant for automated deliverables.
The four-role table for who owes what under Article 50, seven renewal-time questions for your software suppliers, disclosure do's and don'ts, and full citations with key dates.
What You'll Know
Model developers, AI-embedded software, deployer firms, and firms that publish AI output. Who carries which duty, and why your practice is almost certainly a deployer with no marking duty at all.
The monthly pack, the tax return, and the advisory memo are not published, so they are not in scope, no matter how much AI touched them. The worksheet proves it output by output.
Not where the law looks. The watermark flags your most-governed content hardest and misses unattended automation. The matrix shows which outputs need a workflow fix and which just need a story.
The template statement, the by-design variant for automated deliverables, and the do's and don'ts that keep you off both failure modes: blanket labels and unprovable "handcrafted" claims.
Your software's AI features carry their own marking duty, and their answers land in your client-facing output. Seven questions to ask at renewal, not after a client asks you.
Stay Current
Subscribe to The AI Accountant newsletter and get The AI Disclosure Position Kit delivered to your inbox, along with weekly analysis of the AI developments that matter for your practice.
Your Move
The detectors are live and the watermark is already in your tools. The firms that get hurt won't be the ones whose content flags as AI; they'll be the ones whose flag arrives with no story attached. Write yours in one sitting.