On October 22, three of the big coaching bodies, the Association for Coaching, EMCC Global and the International Coaching Federation, signed a framework to work together on AI. There's no new rulebook yet. Each body keeps its own standards, and the plan is to build shared resources for coaches, mentors and supervisors.
What they did say is where they'll focus: client protection, privacy and confidentiality, transparency, appropriate professional boundaries, and accountability for AI use.
You don't need to wait for the shared resources to act on that list. Most of it comes down to one plain question: where does your client's material actually go after a session?
Your Code already covers this
If you're an ICF member or credential holder, the ICF Code of Ethics that took effect on April 1, 2025 already speaks to it. Two standards matter here:
- 2.4: maintain, store and dispose of records, including electronic files and communications, in a way that promotes confidentiality, security and privacy and complies with applicable laws and agreements.
- 2.5: fulfil your ethical and legal obligations to clients, sponsors, colleagues and the public directly and through any technology you use. The Code names artificial intelligence in that list.
So the AI tools you use aren't a side issue. Whatever they store counts as your records.
Most coaches don't know their own data trail
It's easy to think of "client notes" as the doc you type after a session. In practice, a single session can leave copies in five or six places:
- The video platform recording, if you record.
- An AI notetaker that joins the call and saves a transcript and summary.
- A transcription or summary tool you paste the transcript into afterwards.
- A general AI chat tool where you asked for help preparing for the next session.
- Your cloud drive or notes app, where the final notes live.
- Voice memos you recorded on your phone in the car after the call.
- Messages with the client on WhatsApp, email or your scheduling tool.
Each of those has its own storage, its own access settings and its own deletion rules. Some keep data until you delete it. Some may use it to improve their product unless you switch that off. You can't tell a client their information is confidential if you don't know where it all sits.
Do a one-hour records audit
Block an hour this week. Open a blank table with these columns:
| Tool | What it stores | Where | Who can see it | How long it's kept | How to delete it |
|---|
Then go through every tool that touches session content and fill in a row. Use the tool's own settings and help pages, not your memory.
A few questions to ask for each one:
- Does it keep recordings or transcripts, and for how long? If you can set a retention period, set one.
- Is my data used to train or improve the product? Look for the setting. If there's an opt-out, decide on purpose whether to use it.
- Who else can see it? Shared workspaces, assistants, team members, the platform's support staff.
- Can I delete one client's data on request? If a client ends the work and asks you to delete their records, could you actually do it everywhere?
You'll likely find at least one tool you'd forgotten about. That's the point of doing it.
Cut the trail down
Once the table is filled in, shrink it.
- Drop tools you don't need. If two tools both make summaries, keep one.
- Stop pasting full transcripts into general AI chats. If you want help thinking about a client's pattern, describe it without names, employers or details that would identify them.
- Set a deletion rhythm. For example, delete raw recordings once your notes are written, and review stored notes when an engagement ends. Pick what fits your practice and the laws where you work, then stick to it.
- Treat voice memos like notes. If you record reflections on your phone, move them into your notes system and delete the originals.
Fewer copies means fewer places for something to leak, and less to explain.
Tell clients what you actually do
Transparency was on the joint list for a reason. Once you know your data trail, share a short version with clients. Something like:
I use [video tool] for sessions and [notetaker] to create a transcript. Transcripts are deleted after I write my notes. My notes are stored in [tool] and kept for [period] after we finish. I don't put your name or identifying details into general AI tools.
Fill in your real tools and periods. If you already have an AI clause in your coaching agreement, this is the practical detail behind it. If you don't, our earlier piece on writing an AI clause into your coaching agreement covers the consent side.
Sponsors complicate it
If a company pays for the coaching, there may be three parties with views on data: you, the client and the sponsor. Section 2 of the ICF Code already asks for clear agreements about what information is shared and how.
Add AI tools to that conversation. A company may have rules about which recording or AI tools can be used with its staff. Ask before the first session, not after a sponsor's IT team finds your notetaker in the call.
Watch what comes next
The three bodies said they'll develop shared resources, and ICF Converge runs October 28 to 30 in Washington, D.C., according to ICF's site. It's reasonable to expect more guidance on AI over the coming months.
When it arrives, a coach who already has a records table will just need to check it against the new guidance. A coach who doesn't will be starting from zero. An hour now is a cheap way to be in the first group.