Metraipolis › archive › dc_d9483ea08d06
On the Impact of Anonymization on the Performance of Large Language Models — arXiv:2609.11335
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.11335
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T05:27:02.091Z
- SHA-256 of the archived text
d3669031e2adc803996d5986b1e39299bb9ba424083dc1b4ebaa19668e439a89- Stable URL of this page
https://metraipolis.com/documento/dc_d9483ea08d06
How to cite
On the Impact of Anonymization on the Performance of Large Language Models — arXiv:2609.11335 — arXiv (en), https://arxiv.org/abs/2609.11335 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d9483ea08d06 (Metraipolis archive dc_d9483ea08d06, sha256 d3669031e2adc803996d5986b1e39299bb9ba424083dc1b4ebaa19668e439a89)
BibTeX
@misc{dcd9483ea08d06,
title = {On the Impact of Anonymization on the Performance of Large Language Models — arXiv:2609.11335},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.11335},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d9483ea08d06; sha256 d3669031e2adc803996d5986b1e39299bb9ba424083dc1b4ebaa19668e439a89},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.11335. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
As large language models are increasingly deployed in sensitive domains, anonymizing input data to protect personally identifiable information has become a critical practice. However, the impact of this anonymization on model utility is not well understood. This paper presents a systematic empirical study of the trade-off between privacy and performance. We evaluate five prominent language models across eleven diverse benchmarks, comparing their performance on original versus pseudonymized inputs. Our results reveal that while anonymization generally degrades performance, the effect is highly nuanced. We find that more capable models, such as Qwen2.5-72B and GPT-4o mini, suffer the largest performance drops, suggesting a stronger reliance on specific entity information. The impact is also task-dependent: performance on TruthfulQA improves with anonymization, while retrieval-focused tasks like RGB experience a catastrophic decline. Further experiments show that reversible anonymization techniques that preserve entity uniqueness significantly outperform irreversible ones like redaction, and that explicitly prompting models about anonymization offers no discernible benefit. We conclude that anonymization is not a one-size-fits-all solution and must be co-designed with the model and task in mind to balance privacy and utility effectively. Our findings provide a crucial baseline for developing more robust, privacy-aware AI systems. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-11 in cs.AI). Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.11335 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-16T05:17:52.152Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).