Somebody has filed a #GDPR complaint about #OpenAI, because they are unable to correct false personal data that is produced by #ChatGPT.
This should be interesting to watch.
@swaldman@ErikJonker as an american, i was a bit shocked to see this was a privacy issue until i read through it the first time. i’m curious to hear from a european about their thoughts on this
@kellogh@swaldman ...personally i am not sure whether this is the right way to adress this issue. Ofcourse the problem of training LLMs with large pools of unlicensed data, including private/personal data without permission (when it's needed) should be adressed, i only don't know if this is the way...🤔
@kellogh@ErikJonker As a non-expert European: the law applies to "personal information" stored in pretty much any way. Doesn't have to be a database. Could be in a notebook in a filing cabinet.
People have the right to request the information that a company holds on them, and to have any inaccuracies corrected. OpenAI is not able to fulfill either of those requirements, hence appears to me - a non-expert - to be in breach.
@swaldman@kellogh legally I am quite sure most LLMs are a breach of GDPR , because for most the training data is far from transparant and could or does include data related to persons or even personal data 🤔
Add comment