Supervisory authority examines Ebay's AI training with user data
Ebay wants to use users' personal data for the development of AI. The competent authority has received many complaints and is investigating them.
(Image: Mike Knell CC BY-SA 2.0)
The Brandenburg data protection officer Dagmar Hartge has identified a need for clarification regarding new clauses in Ebay's privacy policy. Ebay recently reserved the right to use customers' personal data in order to train and develop its own models and solutions with artificial intelligence (AI). For Hartge, it is currently not sufficiently clear which users' personal information the company wants to use. The exact purposes of the intended AI training are also unclear.
There is also a lack of information on which AI models will be used, complains Hartge, who has "already received numerous complaints" from customers on the matter. The Brandenburg authority is responsible as Ebay has its European headquarters in Kleinmachnow near Potsdam.
The online platform is also unlikely to have specified who may use the data at all and whether these rights are restricted to Group employees. In the opinion of the supervisory authority, the legal requirements regarding the transparency of the planned data processing have not yet been met.
Ebay is still waiting with the AI training
According to the data protection officer, she is "in contact" with eBay Germany. The company has stated that it has not yet started the AI training. The amended privacy policy actually came into force on April 21. Hartge recommends that users who do not wish to donate their information for AI training should object to the intended data processing. Once the training has taken place, this "can no longer be reversed" according to the current state of knowledge.
Videos by heise
Ebay refers to the right to opt out in the privacy policy. The objection can be made via the "Settings for AI development and training" in the respective Ebay account. In March, the company indicated that it wanted to use user data for AI-supported bid creation, AI-generated summaries of product reviews and chat responses in real time. To the extent permitted by law, the aim is also to train, test and validate third-party systems. For example, names, addresses, telephone numbers, email addresses, user names and tax identification numbers are recorded, as well as other data records for sellers. Experts have been debating for some time how purpose limitation and data minimization can be designed in a practical way for the age of AI.
(ds)
