Study: AI users talk about sensitive topics, but are plagued by data protection concerns

More and more AI users in Germany are using chatbots with artificial intelligence as contact persons for personal and confidential topics. At the same time, however, they view the technology providers with considerable mistrust when it comes to data protection. This is the central result of a study by the Swiss software service provider Proton.

According to the representative survey, 68 percent of the chatbot users surveyed in Germany have already discussed at least one sensitive topic with an AI system. Users most often turn to digital assistants when they have questions about mental health. In addition, intimate areas of life such as finances, relationships, work, sexuality, secrets or family conflicts are regularly disclosed to the programs.

Convenience and non-judgment in focus

However, according to the study authors, the turn to artificial intelligence does not mean that the technology will completely replace the social environment. Friends and family continued to be the most important point of contact for sensitive issues. The decisive factor for the use of chatbots is their constant availability around the clock. The fact that the systems react without personal prejudice or social embarrassment also plays an important role.

Concern about commercial advertising profiles prevails

At the same time, the results reveal clear skepticism towards the operators of AI services. Around 40.5 percent of those surveyed said they had little or no confidence that companies would adequately secure confidential data. Concern about commercial exploitation is particularly pronounced: 51 percent of users fear that individual marketing or advertising profiles could be created from the dialogues. The risk that their own statements will be used to train new AI models is a concern for 43 percent of participants.

Demand for data protection-oriented systems

Stricter security guarantees could further increase the openness of users. 71 percent of survey participants said they would prefer to use a chatbot that is specifically geared towards data protection. Almost 44 percent also signaled a greater willingness to share intimate details, as long as it is ensured that the content is not used for model training.

For the survey, Proton surveyed a total of 4,014 active users of AI chatbots in Germany, France, Great Britain and the USA between July 22 and August 2, 2026.