Your records are stored in the UK data-residency region set for your country, and speech is transcribed in memory with the audio discarded as it goes.
Notes are drafted by Anthropic's AI model in the United States, outside your residency region. The consent card reads UK GDPR · audio never stored.
When drafting moves into your region, the consent card will say so.
The drafting model only ever receives transcript text, never audio. Anthropic processes that text only to return the draft, under a contract that forbids other use, and it is not used to train any AI model.