AISPA: User-Centric System Prompt Auditing for Large Language Model Applications

Stanford University

Research official 2 src. ~1 min

Introduces an 8-dimension auditing framework (identity transparency, privacy, manipulation prevention, harm prevention, fairness, etc.) grounded in AI-safety guidelines and human-rights articles, then audits 3,249 system-prompt instructions across 88 commercial AI products.

Why it matters

Empirical finding: 98.9% of audited products include at least one protective instruction, but only 24% cover all eight safety dimensions, and roughly 40% contain at least one instruction that works against user interests.

Importance: 2/5

Notable empirical safety audit, official arxiv/HF source only; 28 HF Daily Papers upvotes, below the 100-upvote bump threshold.

Sources