Post-deployment feedback can serve as a scalable, contextual signal for monitoring and improving generative-AI systems. The study treats it as part of how providers monitor system behavior and refine model-enabled products. The supplied evidence does not establish whether feedback satisfies any post-market or other legal requirement.
Post-deployment feedback from users can be a cost-effective, scalable, and representative means toOct monitor and improve generative AI systems and features. When implemented effectively, giving such 2 feedback can increase users’ engagement with and trust in GenAI systems. Government regulations and industry guidelines call for post-deployment user engagement, but there is little guidance on designing mechanisms that are usable for consumers and provide actionable input for product teams. We conducted a multi-phase study as a collaboration between academic researchers and eBay. Our benchmark evaluation of current industry approaches identified common issues including lack of discoverability, unclear terminology, and inattention to user value. Based on these findings, we[cs.AI] developed best-practice recommendations and designed and tested a prototype feedback-collection tool. The tool aimed to provide users with an efficient, flexible, and positive feedback-giving experience, and provide product teams with rich data on performance and potential problems in a usable format.
Generative AI (GenAI) systems, especially Large Language Models (LLMs), have rapidly moved from research settings into everyday digital products Sun et al. [2024]. GenAI capabilities are now embedded in conversational assistants and other interfaces that support content creation and information work OpenAI [2022], Tahaei et al. [2023a]. Human feedback is central to how these systems are built and improved Hashmati et al. [2025]: it enters through data creation and annotation, supervised fine-tuning and reinforcement learning, and post-deployment interaction and evaluation National Institute of Standards and Technology [2023], of Standards and Technology [2024]. In deployed products,arXiv:2610.02631v1 feedback is therefore not a peripheral interface feature; it is part of how providers monitor system behavior and iteratively refine model-enabled systems National Institute of Standards and Technology [2023], of Standards and Technology [2024]. Figure 1 shows some examples of feedback forms for GenAI products and features.
In the reviewed sample, 8 of 40 GenAI products had no user-feedback mechanism, while 80% had one. Existing mechanisms often adapted generic satisfaction or bug-reporting flows and did not consistently cover GenAI-specific concerns such as bias, offensive content, inaccurate responses, or unrealistic generated images. Participants also had more difficulty finding feedback when entry points were hidden, separated from the generated output, or placed in menus, and they were not always sure what the feedback covered.