Quantization Undoes Alignment: Bias Emergence in Compressed LLMs Across Models and Precision Levels
The paper reveals that quantizing LLMs to lower precision (e.g., 3-bit) causes 6-21% of previously unbiased items to exhibit new stereotypical behaviors, while models' willingness to answer 'unknown' drops by 17.4%. Standard metrics like perplexity barely change, indicating aggregate metrics miss fairness-critical degradation.
Article intelligence
Key points
- Quantization can reintroduce biases even in aligned models.
- 3-bit quantization caused 6-21% of unbiased items to become stereotypical.
- Models' selection of 'unknown' answers declined by 17.4%.
- Standard metrics like perplexity fail to detect fairness degradation.
Why it matters
This matters because quantization can reintroduce biases even in aligned models.
Technical impact
May affect model selection, inference cost, product capability, and evaluation benchmarks.
[2605.15208] Quantization Undoes Alignment: Bias Emergence in Compressed LLMs Across Models and Precision Levels
[Submitted on 2 May 2026]
Title:Quantization Undoes Alignment: Bias Emergence in Compressed LLMs Across Models and Precision Levels
View a PDF of the paper titled Quantization Undoes Alignment: Bias Emergence in Compressed LLMs Across Models and Precision Levels, by Plawan Kumar Rath and 1 other authors
View PDF HTML (experimental)
Abstract:Large Language Models are routinely compressed via post-training quantization to reduce inference costs and memory footprint for cloud and edge deployment, yet the impact of this compression on model quality remains poorly understood. Existing studies typically compare only two conditions (full-precision vs. a single quantized variant), rely on aggregate bias metrics, and evaluate a single model family, making it impossible to distinguish gradual degradation from threshold-dependent safety failures. We conduct a controlled empirical study of three instruction-tuned models (Qwen2.5-7B, Mistral-7B, Phi-3.5-mini) at five precision levels (BF16 through 3-bit) on 12,148 BBQ bias benchmark items across 5 random seeds, totaling 911,100 inference records. Our results reveal that 3-bit quantization causes 6-21% of previously unbiased items to develop new stereotypical behaviors, following a clear dose-response pattern confirmed via logistic regression, while models' willingness to select "unknown" answers declines by 17.4%. Crucially, these item-level changes are invisible to standard quality metrics: perplexity increases by less than 0.5% at 8-bit and under 3% at 4-bit across all three models, yet 2.5-5.6% of items already develop new biases at 4-bit. These findings demonstrate that aggregate evaluation metrics systematically miss fairness-critical degradation, underscoring the need for quality-aware compression protocols that explicitly test for bias emergence before deployment.
Comments: 7 pages, 4 figures, 4 tables. Accepted at IEEE Cloud Summit 2026. This is the author's accepted version; the version of record will appear in IEEE Xplore
Subjects:
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
ACM classes: I.2.7; I.2.6
Cite as: arXiv:2605.15208 [cs.LG]
(or arXiv:2605.15208v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2605.15208
arXiv-issued DOI via DataCite
Submission history
From: Plawan Kumar Rath [view email] [v1] Sat, 2 May 2026 05:41:47 UTC (869 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled Quantization Undoes Alignment: Bias Emergence in Compressed LLMs Across Models and Precision Levels, by Plawan Kumar Rath and 1 other authors
View PDF
HTML (experimental)
TeX Source
view license
Current browse context:
cs.LG
new | recent | 2026-05
Change to browse by:
cs cs.AI
References & Citations
NASA ADS
Google Scholar
Semantic Scholar
Loading...
Data provided by:
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
IArxiv recommender toggle
IArxiv Recommender (What is IArxiv?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)