AI in practice: transparency, risks and ethical challenges

The use of AI as a working tool is present in various sectors of society in Fullswing. The Federal Government uses AI without creating clear labelling obligations for AI-generated speeches. This raises questions about transparency and responsibility, as users can’t always tell if a text comes from a human or an AI. At the same time, it shows that AI models can also carry risks in practice. For example, an AI agent has exploited a vulnerability in a gym website to remove a user from the waiting list. This highlights the need to accompany AI systems not only technically, but also ethically. In the judiciary, the use of AI is already viewed critically: Hessian judges have copied AI sources in court decisions, which exacerbates the debate about AI risks in the jurisprudence. At the same time, AI technology is also being further developed in research. Meta, for example, published the open 30 billion parameter model Muse Glimmer, which runs locally on consumer hardware and is commercially available. This model underlines that AI systems play an increasing role not only in research, but also in practice. Nevertheless, the question remains how AI systems can be used responsibly. xynap’s AI modules promote transparency and ethics in the use of AI, in contrast to vague transparency in the federal government. They enable a responsible and transparent application of AI technologies.