TSR Desk · science · 18 October 2026, 01:00 UTC
In the Guise of Correction: How Framing and Cultural Adaptation Erode LLM Safety on Real-World
- What
- In the Guise of Correction: How Framing and Cultural Adaptation Erode LLM Safety on Real-World Criminal Cases
- Who
- arxiv.org
- When
- 17 October 2026, 04:00 UTC
- Category
- Science
- Primary source
- https://arxiv.org/abs/2610.21789
- What is not known
- This brief does not claim independent replication. Claims that appear only on X and not in the primary source stay unknown.
We introduce GUISE, a benchmark of criminal-assistance scenarios sourced from real-world cases and organised under a taxonomy from the UNODC International Classification of Crime for Statistical Purposes (ICCS). It comes from a paper posted to arXiv on 17 October 2026. Existing safety benchmarks rely largely on synthetic prompts with loosely grounded taxonomies and offer little focused analysis of severe crime categories. Each scenario is rendered under three linguistic framings and evaluated across five distinct cultures. Using our proposed two-tier evaluation framework, which judges both harmfulness and operational uplift, we find that: (i) framing alone drives unsafe compliance, with the mean unsafe compliance rate rising from 28.9% under direct requests to 66.4% under correction seeking (peaking at 97.3%; p < 0.05), as models correcting deliberately distorted details supply executable criminal steps; (ii) culturally adapted prompts elicit significantly more unsafe compliance than literal translations of the same requests in all four non-English languages (+7 to +18 points), a risk that translation-based benchmarks miss; and (iii) unsafe responses from the least-safe models carry strong Anglocentric registers, and the extent of this leakage increases monotonically with response harmfulness ($\rho$ = 0.46 to 0.63 across languages, all p < 0.001). GUISE offers a grounded resource and fine-grained diagnostics for strengthening refusal alignment across languages and cultures. We publish our code and data to facilitate future research.
Why it counts
We introduce GUISE, a benchmark of criminal-assistance scenarios sourced from real-world cases and organised under a taxonomy from the UNODC International Classification of Crime for Statistical Purposes (ICCS).
Sources
Primary source: primary source
What is not known
This brief does not claim independent replication. Claims that appear only on X and not in the primary source stay unknown.
No clip. The article still stands.