Data Protection Officer: Use rules to slow down AI when collecting data

The Federal Commissioner for Data Protection and Freedom of Information, Ulrich Kelber, calls for rules for the collection of data by AI-supported services such as ChatGPT. If personal data is publicly accessible, it would have to be specially protected from the point of view of data protection and should under no circumstances be used for any purpose, Kelber said in an interview with Deutschlandfunk. This applies not only, but especially to artificial intelligence. Technical measures are therefore also required to protect your data on the Internet from being recorded by such systems. Users need clarity that their data is blocked.

Advertisement

Data blocked – or anonymized

With his position, Kelber opposes the attitude of most internet companies, which have been collecting freely accessible data on the internet in bulk and using it to train their AI systems. In the context of AI, the implementation of data protection laws is challenging, Kelber admitted in the “Interview of the Week” at DLF. One difficulty is that the training data is no longer identical to the raw data, but has already been processed in the AI ​​systems. This makes it more difficult to enforce data protection rights, something to delete personal data.

Kelber therefore demanded that the problem be tackled as early as the acquisition of AI training data. For example, such data would have to be pseudonymized or anonymized before AI systems could be trained with it. There should not be a risk that knowledge about a specific person could later be gained from an AI system.

Rules are on their way

The collection of data by AI systems has been under critical observation for some time, and regulatory interventions by some states are already being considered or threatened. For example, the seven largest IT companies recently made a commitment to US President Biden to further develop their AI systems in such a way that they are safe for everyone involved. Your content should be marked as such and checked for security risks.

In the EU, on the other hand, there are plans for particularly strict rules to which AI should be subject: several committees of the EU Parliament envisage that it must follow democratic rules, serve the common good and protect civil rights. The proposals would hit popular AI chatbots like ChatGPT particularly hard, which is why its company OpenAI has already complained about the threat of over-regulation.

Advertisement

(tiw)

Home
#Data #Protection #Officer #rules #slow #collecting #data