arXiv AI

Harm Laundering in GPT Models: Evidence That Gender Discrimination Is Transformed Rather Than Reduced Across Safety-Trained Generations

The paper investigates how safety evaluations for large language models may mask ongoing gender discrimination by transforming harmful content rather than eliminating it, a phenomenon termed "harm laundering." Analyzing 450,000 gender‑directed completions across GPT‑2 to GPT‑5, the authors find that sexual violence content directed at women disappears while men receive more positive representations, with GPT‑5 showing stark disparities such as framing breast cancer as a men’s rights debate. The study introduces a formal test and detection protocol for harm laundering, demonstrating that reduced toxicity scores do not necessarily reflect reduced representational harm.

Hugging Face Trending Papers

Sep 17

Harm Laundering in GPT Models: Evidence That Gender Discrimination Is Transformed Rather Than Reduced Across Safety-Trained Generations

The paper investigates how safety evaluations for large language models may mask ongoing gender discrimination, a phenomenon the authors term "harm laundering." By analyzing 450,000 gender‑directed completions across GPT‑2 to GPT‑5, they show that harmful content directed at women is transformed rather than removed, while men receive more positive representations. The study introduces a formal test and detection protocol for harm laundering, demonstrating that reduced toxicity scores do not necessarily indicate reduced representational harm.

llms safety

arXiv AI

3d ago

Gender bias across LLMs is common and highly heterogeneous

arXiv:2609.38036v2 Announce Type: replace-cross Abstract: Understanding gender biases in large language models (LLMs) is increasingly important as these systems become embedded in decision-support to...

By Edoardo Bolzoni, Valerio Capraro

llms safety

arXiv AI

4d ago

Gender bias across LLMs is common and highly heterogenous

arXiv:2609.38036v1 Announce Type: cross Abstract: Understanding gender biases in large language models (LLMs) is increasingly important as these systems become embedded in decision-support tools with...

By Edoardo Bolzoni, Valerio Capraro

llms safety

arXiv AI

Sep 17

Newer Is Not Fairer: Gender Stereotyping in Text-to-Image AI Across Model Generations

The study evaluates gender representation in 8,000 images generated by four generations of the Stable Diffusion text‑to‑image model across 20 occupations and five prompt templates. It finds that 76.4% of the images depict male subjects, with 57.6% of historically female‑coded occupations also showing male subjects, and that newer model generations do not consistently reduce bias. Compared to U.S. Bureau of Labor Statistics data, the models underrepresent women by 20–46 percentage points, especially in near gender‑balanced fields such as scientists and cleaners.

By Shesh Narayan Gupta, Nik Bear Brown

llms diffusion safety

arXiv AI

Aug 18

HarmProfile: Characterizing Harmful Distributions in Frontier LLMs

arXiv:2608. 14577v1 Announce Type: cross Abstract: Frontier large language models (LLMs) safety evaluation has largely treated harmful generation as an attack outcome rather than as an object of analysis.

By Zhouyuan Ma, Yutao Wu, Hanxun Huang, Xiang Zheng, Xiao Liu, Yixin Cao, Zuxuan Wu, Xingjun Ma, Yu-Gang Jiang

llms benchmarks safety

arXiv AI

Jun 2

The Refusal--Compliance Tradeoff: A Large-Scale Safety Behavior Audit of Large Language Models

arXiv:2605. 05427v2 Announce Type: replace Abstract: Refusal rates are a poor proxy for LLM safety, i.

By Alif Al Hasan, Sumon Biswas

llms benchmarks safety

arXiv AI

Jun 4

AICompanionBench: Benchmarking LLMs-as-Judges for AI Companion Safety

arXiv:2606. 04867v1 Announce Type: new Abstract: As AI companion platforms such as Replika and Character.

By Yanjing Ren, Reza Ebrahimi, TengTeng Ma

llms robotics benchmarks safety

arXiv AI

Sep 17

Linguistic Triggers of Gender and Racial Bias in Open-Weight LLMs Applied to Recruitment

The paper reports the first systematic audit of open‑weight large language models (LLMs) in hiring contexts, examining how job‑posting language influences recruiter and job‑seeker simulations across six models. It finds that agentic language lowers recruiter scores for female candidates while communal language mitigates this effect, and that coded‑exclusion language sharply reduces recruiter scores for non‑White candidates and discourages non‑White personas from applying. The study also identifies the explicit demographic label as the main causal factor and proposes a concrete pre‑deployment audit protocol aligned with EU and U.S. regulatory requirements.

By Kosuke Kitahara, Nobuhiro Yamaguchi

llms rag agents nlp benchmarks safety

arXiv AI

Jun 29

Health-ORSC-Bench: A Benchmark for Measuring Over-Refusal and Safety Completion in Health Context

arXiv:2601. 17642v2 Announce Type: replace Abstract: Safety alignment in Large Language Models is critical for healthcare; however, reliance on binary refusal boundaries often results in over-refusal of benign queries or unsafe compliance with harmful ones.

By Zhihao Zhang, Liting Huang, Guanghao Wu, Preslav Nakov, Heng Ji, Usman Naseem

llms benchmarks safety

arXiv AI

Jun 16

AgentFairBench: Do LLM Agents Discriminate When They Act?

arXiv:2606. 16723v1 Announce Type: new Abstract: Large language model (LLM) agents increasingly take actions (screening applicants, recommending credit, triaging patients), yet fairness for LLMs is still measured by grading answers.

By Triveni Morla, Rohith Reddy Bellibaltu, Manpreet Singh, Manmeet Singh Kapoor

llms agents benchmarks safety

arXiv AI

Sep 16

How Humans and LLMs Read Gender into Gender-Neutral Physical Descriptions

arXiv:2609.16366v1 Announce Type: cross Abstract: When foundation models describe people, recent work in AI fairness, accessibility, and ethics recommends avoiding inferred identity labels (e.g., "sh...

By Yingjia Wan, Lin Lin, Elisa Kreiss

llms safety

arXiv AI

Jun 3

Generating the Modal Worker: A Cross-Model Audit of Race and Gender in LLM-Generated Personas Across 41 Occupations

arXiv:2510. 21011v3 Announce Type: replace-cross Abstract: As generative AI tools are increasingly used to portray people in professional roles, understanding their racial and gender representational biases is critical.

By Ilona van der Linden, Sahana Kumar, Arnav Dixit, Aadi Sudan, Smruthi Danda, David C. Anastasiu, Kai Lukoff

llms safety

Related stories