Privacy-friendly AI training is particularly important in the fields of Artificial Intelligence, Digital Society, and Cybercrime and Cybersecurity. The term describes methods in which artificial intelligence (AI) is „trained“ in such a way that personal data protection is always maintained.
For an AI to learn, large datasets are often used, such as images, texts, or customer data. However, in privacy-friendly AI training, this data is processed in such a way that conclusions about individuals are impossible. This means that personal information like names, addresses, or sensitive details are removed, anonymised, or not processed at all.
A clear example: a supermarket wants to use AI to better understand shopping habits. However, instead of using individual customer data like their name or account number, the supermarket processes the data before training the AI in such a way that personal identification is not possible – for example, by summarising data or omitting sensitive information.
Privacy-friendly AI training ensures that companies can develop innovative solutions without violating data protection laws or jeopardising customer trust.
Deutsch 
English (UK)