The forgotten call: what your AI-recording policy is likely leaving out
Dr John Yardley, founder and CEO at Threads, shares key action points for ensuring secure AI-enabled call recording amid increasing litigation against AI notetakers
There’s now a growing string of lawsuits against AI meeting notetakers over whether every participant on a call properly consented to being recorded. Nothing has been decided yet, and the details of the cases aren’t for us to comment on. What matters for practice managers is the pattern: firms are being pushed, fast, to get consent and disclosure right for AI tools that join calls and meetings.
These are important decisions to get right, but I’m finding myself wondering if firms are asking all the questions they need to.
The channel nobody’s policy covers
Every one of those cases, and most of the internal policy work firms are doing in response, is about video meetings because that’s where AI notetakers currently operate. They join a scheduled call on a platform built to let a third-party tool in.
A phone call doesn’t offer that entry point. There’s no link for a bot to join. So, a firm can build a solid consent process for video meetings including: a disclosure at the start of the call, a record of who agreed, a way to remove the notetaker, and still have no equivalent process for the callback made from a fee earner’s desk, or the update given over the office line between meetings.
This is a key distinction and arguably the bigger area by volume of conversations: for most practices, phone calls still account for a large share of substantive client contact, including urgent queries and time-sensitive advice.
Why it matters when something goes wrong
The video meeting question firms are answering right now is: ‘did we get consent, and can we prove it?’ The equivalent question for phone calls is different and mostly unasked: ‘is there any record of this conversation at all, beyond memory?’
That distinction matters a lot for complaints handling. Disputes about advice given on a call rarely turn on some abstract recording question; they turn on two people recalling the same conversation differently, with nothing to settle it. A verbatim record is designed to solve that challenge.
If a firm has done the work to make its video meetings defensible but hasn’t asked the same question of its phone calls, it’s solved a visible problem and left the larger one untouched.
What’s worth asking this week
This isn’t a legal question so much as an operational one, and it doesn’t require changing the behaviour of your staff to start addressing it. Here are some key considerations:
- Map the channel split. Ask fee earners, roughly, what share of substantive client contact happens by phone versus video or email. Most firms haven’t measured this and are surprised by the answer.
- Check what is currently recorded on calls. For calls made through the office phone system, what is captured today? Recordings, transcription, notes saved to case files. Or does it depend entirely on the individual writing something down afterwards?
- Does your firm have a plan for all the types of calls that happen? Calls made from the office, via a mobile or via a meeting tool.
- Ask where the audit trail actually breaks. If a complaint arrived tomorrow about advice given on a call, would there be a record to check or just recollection?
- Crucially: Is the call automatically recorded, transcribed and filed? Any manual system quickly breaks down. Participants forget, think the call is not worth recording or file it somewhere it cannot be found when needed.
A quieter question to consider
Most firms writing AI policy right now are focused on the notetaker joining the meeting room, and the current wave of litigation shows why that focus makes sense. But it’s worth putting a second, quieter question to your own practice alongside it: Of the client conversations that happen in a week, how many are on the phone, and how many of those currently produce any record beyond memory, if they do produce a record, what happens with it?
From my experience, it is the legal profession that seems to have the most to gain by efficiently structuring their operational AI and call transcription use.
www.threads.cloud
