AI Accountability
The question of who is liable for due diligence and risk assessment when something goes wrong with AI use in a production, and at what stage of the production this should be considered.
Key questions raised (from the paper)
- Who is liable for due diligence and risk assessment, especially if something goes wrong? The prompt engineer? The production company? The technical team? The AI model provider or the creator platform?
- At what stage of the production should this be considered?
- What exactly are the guarantees of legal indemnities (contractual safeguards against copyright infringement lawsuits from generative AI output or training data)? Under what conditions do such indemnities become void?
FRAMES context
- Accountability is one of the responsible-AI themes (see responsible-ai) alongside copyright, transparency and disclosure.
- The paper’s closing argument (Tim Deussen, XRBB) frames the hard problem as “proof”: regulation asks a production to account for itself at the level of the finished work, tools mark at the level of a single output, and authorship/rights live at the level of the individual asset - nothing today carries a claim across all three, so people carry it instead.