From Alignment to Access Control: A Framework for GenAI Policy Enforcement
A framework maps generative AI policy enforcement from model alignment to access control.
Nathalie Baracaldo surveys policy enforcement for generative AI chat and agent applications, arguing that inconsistent meanings of policy produce siloed controls inadequate for compliance. The paper proposes a method to analyze existing enforcement approaches and issues recommendations for the community. It is a companion to the author's USENIX Security 2026 Enigma talk on a unified view from model alignment to access control. It notes that safety mechanisms have lagged rapid capability growth and that real incidents have already occurred.
- Frames GenAI policy from alignment through access control
- Argues inconsistent policy definitions create siloed controls
- Companion to a USENIX Security 2026 Enigma talk
- Recommends systematic analysis of enforcement approaches
From the original article
1 min read at arxiv.orgGenerative AI (GenAI) applications have flourished enabling users to chat with large language models, and to create agents to act on their behalf for a variety of tasks. The pace of development of capabilities in this field is incredibly fast with security and safety taking a back seat. Unfortunately, the slower pace at which security and safety mechanisms have evolved has led to real incidents.
By Nathalie Baracaldo · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.