This paper introduces a framework to audit system prompts in AI applications, examining how developers design and use these prompts to govern the behaviors of foundation models. Practitioners should care because the lack of transparency and accountability in system prompts can erode trust in AI systems.
Firehose
Filtered to tagged “trust and accountability in AI systems” · clear filters
Browse: People · Companies · Papers · Podcasts · Hacker News · Deep dives
Browse by tag
artificial intelligence 57continual learning 24agentic coding 21open-weight models 20AI 16AI agents 13reinforcement learning 11cybersecurity 9AI safety 8finance 8language models 7large language models 7open-source 7productivity 7deep learning 6machine learning 6natural language processing 6Reinforcement learning 6tech 6Databricks 5robotics 5software development 5Agentic AI 4benchmarking 4Diffusion models 4multi-agent systems 4Recursive self-improvement 4world models 4AI ethics 3AI infrastructure 3