# PolicyShiftGuard: Benchmarking and Improving Policy-Adaptive Image Guardrails

Page: https://stenobird.com/podcast/daily-paper-cast-7079649/policyshiftguard-benchmarking-and-improving-policy-adaptive-image-guardrails
Text version: https://stenobird.com/podcast/daily-paper-cast-7079649/policyshiftguard-benchmarking-and-improving-policy-adaptive-image-guardrails.md
Podcast: [Daily Paper Cast](https://stenobird.com/podcast/daily-paper-cast-7079649)
Published: 2026-07-17T03:36:53+00:00
Episode link: https://share.transistor.fm/s/066736ea
Audio file: https://media.transistor.fm/066736ea/000c5254.mp3
Processing state: not_requested
JSON: https://stenobird.com/v1/public/podcasts/daily-paper-cast-7079649/episodes/policyshiftguard-benchmarking-and-improving-policy-adaptive-image-guardrails
Duration seconds: 1182

## Resource

🤗 Upvotes: 30 | cs.CV, cs.AI, cs.CL Authors: Mingyang Song, Luxin Xu, Haoyu Sun, Minzhou Pan, Yu Cheng, Bo Li Title: PolicyShiftGuard: Benchmarking and Improving Policy-Adaptive Image Guardrails Arxiv: http://arxiv.org/abs/2607.05910v1 Abstract: Image guardrails are typically trained and evaluated under a fixed safety policy, implicitly treating safety as an intrinsic property of an image. Real deployments are different: the same image may be allowed in one product, restricted in another, and newly disallowed when a policy boundary changes. We study policy-adaptive image guardrailing, where a model must decide whether an image violates the currently supplied policy and generalize to held-out policy definitions. We introduce PolicyShiftBench, a comprehensive benchmark with 2,000 policy-discriminative instances over 265 images, where each image is paired with 7.55 policy-conditioned prompts on average to test whether models adapt to the active policy rather than relying on image-level safety priors. We then propose PolicyShiftGuard, a compact policy-conditioned guardrail trained with a two-stage training recipe that combines Randomized Policy SFT (RP-SFT) with Boundary-Pair Policy Adaptation (BP-Adapt). BP-Adapt trains matched prompts for the same image and risk category using standard label supervision and a pairwise comparison loss that separates blocking policies from passing policies. Experiments show that existing VLMs and specialized guardrails remain brittle under policy shifts, while PolicyShiftGuard substantially improves policy-sensitive performance. The 7B model achieves SOTA performance of 76.9 Avg. F1 and 72.1 Avg. PSS on PolicyShiftBench, transfers well to UnSafeBench and SafeEditBench, and improves the latency-performance trade-off with a concise output fo…

## Actions

- request_transcript: `POST https://stenobird.com/v1/public/podcasts/daily-paper-cast-7079649/episodes/policyshiftguard-benchmarking-and-improving-policy-adaptive-image-guardrails/transcription-requests` — Idempotently request low-priority transcript generation for this episode.
- read_markdown: `GET https://stenobird.com/podcast/daily-paper-cast-7079649/policyshiftguard-benchmarking-and-improving-policy-adaptive-image-guardrails.md` — Read the agent-friendly Markdown representation of this episode resource.

A page view does not enqueue transcription. Agents should invoke `request_transcript` explicitly when they need this episode processed.

## Transcript

Full transcripts are not published on public pages unless there is a clear rights basis.