SARs generated by AI

Posted by: Sarah - Posted on:

The ICO have released some guidance to support organisations when the receive AI generated SARs: Subject access requests generated by artificial intelligence | ICO.

Some key points are:

A SAR is not invalid just because it was generated or assisted by AI. Organisations must treat AI-assisted SARs in the same way as any other SAR and respond within the usual timescales. Lengthy, formal language, legal references or drafting errors do not invalidate a request if it is clear the individual is seeking access to their personal information.

Existing SAR rules still apply. Organisations can ask for clarification where a request is unclear, carry out only a reasonable and proportionate search, extend the timeframe for genuinely complex requests, and consider whether a request is manifestly unfounded or excessive where justified. The use of AI alone does not make a request complex, excessive or unfounded.

Verify identity and authority carefully. AI-generated requests may contain inaccuracies or false information, and there is a risk of fraudulent use of AI to create convincing fake documents. Organisations should verify identities, contact details and third-party authority where there are concerns.

Train staff and focus on the substance of the request. Staff should not be intimidated by lengthy or legally complex AI-generated correspondence. The key question remains: what personal information is the individual requesting, and what is a reasonable and proportionate way to respond?