Take Note: Cyber-Risks With AI Notetakers

Transcription applications are joining your online meetings, and the piece notes they bring risks to security, operations, reputation and even strategy, whether the tools are official company apps or not. It highlights that notetakers running outside enterprise systems are especially risky, with many vendors lacking basic cybersecurity maturity such as SOC 2, GDPR alignment and strong encryption, and it cites a case where one enterprise saw about 800 notetaker accounts appear in 90 days through invite sprawl.

The article points to a 2025 class action, Brewer v. Otter[.]ai, where recordings were allegedly used to train models and consent issues were disputed, illustrating how transcripts can surface in litigation. It warns that transcripts can end up in third-party systems that not all departments have reviewed or contracted for, and notes the fragility created when vendors sunset or are acquired, leaving data and organisations stranded.

Practical guidance includes establishing an approval process for AI notetaking, specifying who can activate tools and how transcripts are stored, retained, and accessed, and ensuring human review for accuracy and context.