Overview
- A TÜV–Forsa survey finds 65% in Germany now use generative AI, with 45% using it at least weekly and 72% employing it for research; 85% have tried ChatGPT, ahead of Gemini (33%) and Copilot (26%).
- Independent checks report persistent inaccuracies: the EBU found roughly 45% of chatbot answers contained significant errors, and Newsguard says wrong-answer rates rose from 18% to 35% year over year.
- Workers who train models are advising caution, with t3n citing Guardian interviews describing poor instructions, rushed deadlines and low-quality data that undermine chatbot reliability.
- Workplace guidance cited by Chip.de quotes labor lawyer Nathalie Oberthür: employers bear responsibility if they require AI use, whereas employees are accountable when they choose it independently and must verify outputs and protect data.
- Risk perceptions remain high and regulation is widely favored: about half fear data misuse and deepfakes, 83% support tighter rules, yet only 32% have heard of the EU AI Act; Edelman reports 42% in Germany oppose growing AI use, higher shares in the US and UK and lower in China and Brazil.