Police Data Analysis: GFF Warns Against Palantir Use
Constanze Kurz and Franziska Görlitz warn against the use of Palantir by the police. Fundamental rights are at risk.
Bavaria paved the way for Palantir.
(Image: Realfiction)
According to the Society for Freedom Rights (GFF), the use of automated data analysis by the police is increasingly targeting uninvolved individuals. "People end up in police databases quite quickly," said Franziska Görlitz of the GFF. This includes whistleblowers and witnesses. If this data were merged and analyzed, "people could be targeted solely through such a tool."
Complex analysis systems like Palantir Gotham are particularly problematic. "This is not a data matching capability, but rather extremely complex predictive policing or prediction and analysis systems," said Görlitz. The traceability of the results is lost: "Then the question arises at the end: Did the computer arrive at the conclusion that there is a connection or that the person could be dangerous?"
Bavaria as a Door Opener
Bavaria was primarily highlighted critically in this context. There, the state government under Minister-President Markus Söder took a pioneering role in the use of Palantir and quickly created facts – including tests with real data, even before a viable legal basis existed. Kurz spoke of an "interesting legal understanding" by the "sausage king" – referring to Bavarian Minister-President Markus Söder. Bavaria thus opened a door through which other federal states could also more easily enter, for example, through framework agreements without having to conduct their own tenders.
They really had real data from actual people from Bavaria and other federal states for months. Many police internal databases are also accessed if you don't have much to do with the federal state. Not even the Bavarian Data Protection Officer Thomas Petri was informed. He only learned about it from the press.
The deeper the analysis, the stricter the requirements
Although the Federal Constitutional Court clarified in 2023 that automated data analyses are fundamentally possible, it also set strict standards. "The more extensive the analysis, the stricter the requirements must be," Görlitz emphasized. In Hesse, North Rhine-Westphalia, and Bavaria, these requirements are still not being met from the GFF's perspective, which is why several constitutional complaints are pending.
Criticism is also directed at the growing dependence on private providers. "If you become dependent on software tools from other countries or private companies, there is a risk that you will then need them for your own work," warned Görlitz. Transitional solutions quickly become permanent solutions: "Once you have the tool and spend a lot of money on it, how much incentive is there to develop your own?" Prices could also increase rapidly due to dependencies.
From the GFF's perspective, politicians massively underestimate the implications. "Massive data analyses, especially through artificial intelligence, are error-prone and lead to discriminatory results – they are therefore a great danger to fundamental rights."
Videos by heise
Finally, Kurz referred to research by the Swiss magazine Republik, which showed in detail how Palantir "worked its way into authorities over years" and what technological, geopolitical, legal, and ethical risks are associated with it. A publicly accessible report by the Swiss Army was particularly highlighted, whose experts explicitly recommend "refraining from using solutions from the company Palantir."
Palantir feels misunderstood
Also remarkable is Palantir's reaction: For the first time, the company responded to the criticism in a blog post, portraying itself as "misunderstood" and offering discussions to selected actors – "with NDA," as Konstanze emphasized. At the same time, Palantir also actively approached state and federal interior ministries in the summer to once again promote its products. For Konstanze, this shows: The public debate is having an effect – and at least forces the company out of its previous communication refusal in some areas.
(mack)