AI Notetaker Recordkeeping
The advisor left the client call with a tidy summary and three follow-up items. The notes looked accurate, the action items were clear, and the client received a recap email within minutes. But the compliance team did not know where the summary or transcript was stored.AI meeting assistants can join calls, transcribe conversations, summarize discussions, and draft follow-up messages. They can save time, but they also create a practical recordkeeping question: Which outputs must the firm retain, and can it find them when needed?
One meeting, several outputs
An AI-assisted call might produce an audio recording, a transcript, a summary, action items, a draft email, and the final message sent to the client. Those outputs may be stored in different places: the meeting vendor’s system, a notetaker account, a CRM, or an employee’s personal account.
Each output serves a different purpose. The firm needs to know what its tools create, where those materials go, and which ones its recordkeeping policies cover. It should also distinguish an AI-generated draft from the version an employee reviewed and sent.
What regulators have said so far
Existing requirements still apply when firms use AI. In its 2026 Annual Regulatory Oversight Report, FINRA says its rules are intended to be technology neutral and that using generative AI can implicate supervision, communications, recordkeeping, and fair-dealing requirements for member firms. FINRA also identifies summarization and information extraction as its members’ most common generative AI use case. finra.org
For investment advisers, Rule 204-2 requires retention of specified written communications received and sent, including those relating to advice or recommendations. A client recap email containing advice may fall within that requirement. Whether an unsent internal AI summary or a transcript must be retained requires a separate assessment of its content, use, and any other applicable obligations. ecfr.gov
That assessment is easier to make as part of a firmwide process than after employees have begun creating and storing meeting outputs across multiple accounts.
Address recording consent before the call
Recordkeeping is not the only issue. Firms also need to assess recording and privacy requirements before an AI notetaker joins a meeting.
A proposed class action filed against Otter.ai in 2025 alleges that its notetaker recorded and processed meeting conversations without the consent of all participants. In an August 2026 order, the court allowed several federal and state privacy claims to proceed while dismissing others, some with permission to amend. The court had not decided whether the allegations were true. docs.justia.com
The case is a reminder to check the consent rules that apply to a meeting and to make sure participants receive appropriate notice. Firms should also understand what the notetaker vendor stores and how it uses meeting data.
For law firms, the American Bar Association’s Formal Opinion 512 discusses lawyers’ duties when using generative AI, including protecting client information and communicating with clients when required. americanbar.org
Questions to answer before the next client call
- Which AI meeting tools are employees using, including tools they adopted individually?
- What does each tool create: recordings, transcripts, summaries, action items, or drafts?
- Which outputs must the firm retain under its applicable rules and policies?
- Where is each output stored, and can the firm search and export it?
- Who reviews AI-drafted follow-ups before they are sent?
- How are meeting participants notified, and how is any required consent obtained?
- How long does the vendor keep the data, and can it use the data to train its models?
Set a policy and apply it consistently
A firm may have a sound approach for its approved AI meeting tool but little visibility into one connected through an employee’s personal account. Start by identifying the tools in use. Then decide which outputs the firm needs to retain, how employees should review AI-generated material, and where retained records will be kept.
An AI summary can be a useful account of a client conversation, but it can also omit context or contain errors. A consistent process helps the firm preserve records when required, protect client information, and know which version was actually sent or relied upon.
