Using AI Safely in Companies

Starting in September 2026, the consortium will receive €1.29 million in funding over three years from the Federal Ministry of Research, Technology and Space as part of the “Privacy Platform” funding initiative. From the Research Center for Trustworthy Data Science and Security, Dr. Steffen Becker (Ruhr University Bochum), Dr. Greta Ontrup, and Dr. Magdalena Wischnewski (both from the University of Duisburg-Essen) are involved.

Balancing efficiency and data protection
“Many companies have valuable data that, until now, hardly anyone has been able to use systematically. Artificial intelligence is fundamentally changing that,” says project coordinator Dr. Steffen Becker from the Research Center Trustworthy Data Science and Security. Whether it is answering customer inquiries, summarizing contracts, or taking meeting minutes, AI applications can make many everyday tasks easier and faster – and in doing so, they almost always access internal documents as well as customer or employee data. “We are therefore always balancing the efficiency gains offered by AI against the need to protect sensitive data,” says Steffen Becker.

Technical solutions
The research team will first take a close look at how AI is actually being used in companies before developing solutions for its secure deployment. “Once data has been transferred to an AI service, it is almost impossible to fully understand from the outside what happens to it. We therefore want to systematically examine the actual data protection risks associated with commonly used services – and provide companies with tools that prevent sensitive data from being transferred without proper checks in the first place,” Steffen Becker explains. The researchers are considering, for example, filtering solutions that identify confidential information before it leaves the company and either warn users or block its transmission.

The right balance of trust and skepticism
Researchers at Witten/Herdecke University are examining AI-related processes within companies from an organizational psychology perspective. How, for example, do company management teams make decisions? What attitudes do works councils have toward AI?

Researchers at the University of Duisburg-Essen are focusing on employees and contribute the individual psychology perspective. “What matters is that people neither trust AI blindly nor underestimate what it can do,” says Dr. Magdalena Wischnewski. “We are therefore investigating what ideas employees have about how AI works, where misconceptions arise, and how an appropriate level of trust can be fostered,” adds Dr. Greta Ontrup. Both researchers lead Young Investigator Groups at the Research Center Trustworthy Data Science and Security. Among other things, the researchers will develop training formats based on these findings. “We are also addressing the legal aspects through a legal opinion, particularly the interplay between data protection, the AI Act, and labor law,” says Steffen Becker.