Metraipolis › archive › dc_258cf897147a

Anchoring Bias: A Persistent Fairness Backdoor Attack against MLLMs under Continual Learning — arXiv:2608.21577

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.21577
License (as archived): CC0-1.0
Archived: 2026-10-18T07:42:19.597Z
SHA-256 of the archived text: 53f6c04ba6ceae198b9febd7e6da35a3919bfe44e8eccd9d34bdfb0e03d16804
Stable URL of this page: https://metraipolis.com/documento/dc_258cf897147a

How to cite

Anchoring Bias: A Persistent Fairness Backdoor Attack against MLLMs under Continual Learning — arXiv:2608.21577 — arXiv (en), https://arxiv.org/abs/2608.21577 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_258cf897147a (Metraipolis archive dc_258cf897147a, sha256 53f6c04ba6ceae198b9febd7e6da35a3919bfe44e8eccd9d34bdfb0e03d16804)

BibTeX

@misc{dc258cf897147a,
  title = {Anchoring Bias: A Persistent Fairness Backdoor Attack against MLLMs under Continual Learning — arXiv:2608.21577},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.21577},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_258cf897147a; sha256 53f6c04ba6ceae198b9febd7e6da35a3919bfe44e8eccd9d34bdfb0e03d16804},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.21577. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Multimodal Large Language Models (MLLMs) are increasingly deployed in high-stakes domains where fairness is a critical safety requirement. In practice, these models are continually updated through continual learning (CL) to adapt to evolving tasks and data distributions. Prior work has shown that backdoor attacks can manipulate MLLM responses through hidden triggers, but naively implanted backdoors degrade as models undergo subsequent updates of CL. Although fairness has emerged as a central concern for MLLM deployment, whether backdoor-induced fairness violations can survive CL remains unexplored, leaving two critical questions unanswered: (1) whether a backdoor can reliably induce fairness violations in MLLMs, and (2) whether such fairness-targeted backdoors can persist through continual learning. We bridge this gap by proposing Persistent Fairness Backdoor Attack (PFBA) to inject persistent and group-specific discrimination into MLLMs. Specifically, PFBA achieves this through two novel mechanisms. The Latent Space Fairness Reinforcement reshapes the model's deep feature geometry by anchoring privileged-group representations to preserve utility while repelling and clustering targeted-group representations to sustain discrimination, and the Continual Learning Simulation iteratively optimizes the trigger against simulated parameter drift to ensure backdoor persistence across future updates. Extensive experiments demonstrate that PFBA induces severe fairness disparities that persist across continual learning rounds, evading standard backdoor defenses. The data and code are publicly available at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-25 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.21577 — 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 2026-10-18T07:37:52.821Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).