Open-weight genome language model safeguards: Assessing robustness via adversarial fine-tuning

• URL: http://arxiv.org/abs/2511.19299v1
• Date: Mon, 24 Nov 2025 16:46:44 GMT
• Title: Open-weight genome language model safeguards: Assessing robustness via adversarial fine-tuning
• Abstract summary: We evaluate a state-of-the-art gLM, Evo 2, and perform fine-tuning using sequences from 110 harmful human-infecting viruses.<n>We highlight the need for safety frameworks for gLMs and outline further work needed to enable the safe deployment of gLMs.
• Score: 5.053164256829132
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Novel deep learning architectures are increasingly being applied to biological data, including genetic sequences. These models, referred to as genomic language mod- els (gLMs), have demonstrated impressive predictive and generative capabilities, raising concerns that such models may also enable misuse, for instance via the generation of genomes for human-infecting viruses. These concerns have catalyzed calls for risk mitigation measures. The de facto mitigation of choice is filtering of pretraining data (i.e., removing viral genomic sequences from training datasets) in order to limit gLM performance on virus-related tasks. However, it is not currently known how robust this approach is for securing open-source models that can be fine-tuned using sensitive pathogen data. Here, we evaluate a state-of-the-art gLM, Evo 2, and perform fine-tuning using sequences from 110 harmful human-infecting viruses to assess the rescue of misuse-relevant predictive capabilities. The fine- tuned model exhibited reduced perplexity on unseen viral sequences relative to 1) the pretrained model and 2) a version fine-tuned on bacteriophage sequences. The model fine-tuned on human-infecting viruses also identified immune escape variants from SARS-CoV-2 (achieving an AUROC of 0.6), despite having no expo- sure to SARS-CoV-2 sequences during fine-tuning. This work demonstrates that data exclusion might be circumvented by fine-tuning approaches that can, to some degree, rescue misuse-relevant capabilities of gLMs. We highlight the need for safety frameworks for gLMs and outline further work needed on evaluations and mitigation measures to enable the safe deployment of gLMs.

Related papers

• An Early Warning of Emerging Biosecurity Risks in Frontier LLMs [62.01705276160883]
  We develop Intern-BioBreaker, a specialized bio-red-teaming model, together with an integrated computational-to-physical framework.<n>Within this framework, Intern-BioBreaker generates targeted jailbreak prompts to test whether aligned models can be induced to provide operational guidance.<n>Our evaluation reveals a concerning gap between text-level safeguards and the risks posed by capable scientific models.
  arXiv  Detail & Related papers  (2026-07-20T15:26:23Z)
• ViroBench: Benchmarking Nucleotide Foundation Models on Viral Genomics Tasks [86.89727311669937]
  We introduce ViroBench, the first comprehensive and large-scale benchmark specifically designed for nucleotide foundation models (NFMs)<n>ViroBench evaluates models across two critical dimensions: biological understanding and latent biosecurity risk, covering 18 diverse scenarios within 4 task types.<n>ViroBench provides interpretable, diagnostic evaluations and a reproducible measurement framework for future research on viral nucleotide foundation models.
  arXiv  Detail & Related papers  (2026-05-25T03:31:46Z)
• From Data to Behavior: Predicting Unintended Model Behaviors Before Training [78.37660873165284]
  We introduce Data2Behavior, a new task for predicting unintended model behaviors prior to training.<n>We also propose Manipulating Data Features (MDF), a lightweight approach that summarizes candidate data through their mean representations.<n>Experiments on Qwen3-14B, Qwen2.5-32B-Instruct, and Gemma-3-12b-it confirm that MDF can anticipate unintended behaviors and provide insight into pre-training vulnerabilities.
  arXiv  Detail & Related papers  (2026-02-04T16:37:17Z)
• Biosecurity-Aware AI: Agentic Risk Auditing of Soft Prompt Attacks on ESM-Based Variant Predictors [4.781986758380065]
  We introduce the Secure Agentic Genomic Evaluator (SAGE), an agentic framework for auditing the adversarial vulnerabilities of GFMs.<n>Using SAGE, we find that even state-of-the-art GFMs like ESM2 are sensitive to targeted soft prompt attacks.
  arXiv  Detail & Related papers  (2025-12-19T00:51:11Z)
• Adaptive Defense against Harmful Fine-Tuning for Large Language Models via Bayesian Data Scheduler [67.24175911858312]
  Harmful fine-tuning poses critical safety risks to fine-tuning-as-a-service for large language models.<n>Bayesian Data Scheduler (BDS) is an adaptive tuning-stage defense strategy with no need for attack simulation.<n>BDS learns the posterior distribution of each data point's safety attribute, conditioned on the fine-tuning and alignment datasets.
  arXiv  Detail & Related papers  (2025-10-31T04:49:37Z)
• SafeGenes: Evaluating the Adversarial Robustness of Genomic Foundation Models [8.019763193322298]
  We propose SafeGenes: a framework for Secure analysis of genomic foundation models.<n>We assess the adversarial vulnerabilities of GFMs using two approaches: the Fast Gradient Sign Method and a soft prompt attack.<n>Targeted soft prompt attacks led to substantial performance degradation, even in large models such as ESM1b and ESM1v.
  arXiv  Detail & Related papers  (2025-06-01T03:54:03Z)
• Backdoor Cleaning without External Guidance in MLLM Fine-tuning [76.82121084745785]
  Believe Your Eyes (BYE) is a data filtering framework that leverages attention entropy patterns as self-supervised signals to identify and filter backdoor samples.<n>It achieves near-zero attack success rates while maintaining clean-task performance.
  arXiv  Detail & Related papers  (2025-05-22T17:11:58Z)
• HR-VILAGE-3K3M: A Human Respiratory Viral Immunization Longitudinal Gene Expression Dataset for Systems Immunity [8.64940622146001]
  Human Respiratory Viral Immunization LongitudinAl Gene Expression (HR-VILAGE-3K3M) repository integrates 14,136 RNA-seq profiles from 3,178 subjects across 66 studies encompassing over 2.56 million cells.<n>HR-VILAGE-3K3M is the largest longitudinal transcriptomic resource for human respiratory viral immunization.
  arXiv  Detail & Related papers  (2025-05-19T19:37:49Z)
• Exploration of Hepatitis B Virus Infection Dynamics through Physics-Informed Deep Learning Approach [0.0]
  In recent years, Physics-Informed Neural Networks (PINNs) have emerged as a promising framework that can capture the dynamics of viral infection.<n>Standard PINNs are effective in simulating disease dynamics through forward modeling but often face challenges in estimating key biological parameters.<n>DINNs offer a more robust approach to parameter estimation tasks.
  arXiv  Detail & Related papers  (2025-03-12T20:02:31Z)
• Neuromorphic Spiking Neural Network Based Classification of COVID-19 Spike Sequences [4.497217246897902]
  We propose a neural network-based (NN) mechanism to perform an efficient analysis of the SARS-CoV-2 data.<n>In this paper, we introduce a pipeline that first converts the spike protein sequences into a fixed-length numerical representation and then uses Neuromorphic Spiking Neural Network to classify those sequences.
  arXiv  Detail & Related papers  (2024-12-19T10:26:31Z)
• Model Inversion Attacks Through Target-Specific Conditional Diffusion Models [54.69008212790426]
  Model inversion attacks (MIAs) aim to reconstruct private images from a target classifier's training set, thereby raising privacy concerns in AI applications. Previous GAN-based MIAs tend to suffer from inferior generative fidelity due to GAN's inherent flaws and biased optimization within latent space. We propose Diffusion-based Model Inversion (Diff-MI) attacks to alleviate these issues.
  arXiv  Detail & Related papers  (2024-07-16T06:38:49Z)
• Low-rank finetuning for LLMs: A fairness perspective [54.13240282850982]
  Low-rank approximation techniques have become the de facto standard for fine-tuning Large Language Models. This paper investigates the effectiveness of these methods in capturing the shift of fine-tuning datasets from the initial pre-trained data distribution. We show that low-rank fine-tuning inadvertently preserves undesirable biases and toxic behaviors.
  arXiv  Detail & Related papers  (2024-05-28T20:43:53Z)
• Benchmarking Machine Learning Robustness in Covid-19 Genome Sequence Classification [109.81283748940696]
  We introduce several ways to perturb SARS-CoV-2 genome sequences to mimic the error profiles of common sequencing platforms such as Illumina and PacBio. We show that some simulation-based approaches are more robust (and accurate) than others for specific embedding methods to certain adversarial attacks to the input sequences.
  arXiv  Detail & Related papers  (2022-07-18T19:16:56Z)
• UNITE: Uncertainty-based Health Risk Prediction Leveraging Multi-sourced Data [81.00385374948125]
  We present UNcertaInTy-based hEalth risk prediction (UNITE) model. UNITE provides accurate disease risk prediction and uncertainty estimation leveraging multi-sourced health data. We evaluate UNITE on real-world disease risk prediction tasks: nonalcoholic fatty liver disease (NASH) and Alzheimer's disease (AD) UNITE achieves up to 0.841 in F1 score for AD detection, up to 0.609 in PR-AUC for NASH detection, and outperforms various state-of-the-art baselines by up to $19%$ over the best baseline.
  arXiv  Detail & Related papers  (2020-10-22T02:28:11Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.