NeurIPS 2025 Past Safety & alignmentGenerative modelsPrivacy & security
NeurIPS 2025 Workshop on Biosecurity Safeguards for Generative AI
BioSafe GenAI 2025
- Submission deadline
- Sep 15, 2025, 11:59 UTC imported from OpenReview — check the website for extensions
- Submission portal
- OpenReview
- Notes
- Auto-imported from the OpenReview venue record on 2026-06-10 — please verify and enrich (topics are keyword-guessed).
Accepted papers (38)
Fetched from OpenReview (v2) on 2026-06-10.
-
A Biosecurity Agent for Lifecycle LLM Biosecurity Alignment
-
ABC-Bench: An Agentic Bio-Capabilities Benchmark for Biosecurity
-
Agentic BAIM-LLM Evaluation (ABLE): Benchmarking LLM Use of Protein Design Tools
-
AI Bioweapons and the Failure of Inference-Time Filters
-
Behavioral Red Teaming: Investigating Future Biosecurity Risk from Agentic AI and De Novo Sequence Design
-
Benchmarking Biosafety in Generative Protein Design: A Stress-Test Framework for Binder Models
-
Benchmarking diffusion models for predicting perturbed cellular responses
-
Benchmarking Mitigations Against Covert Misuse
-
Best Practices for Biorisk Evaluations on Open-Weight Bio-Foundation Models
-
Biorisk-Shift: Converting AI Vulnerabilities into Biological Threat Vectors
-
Biosecurity-Aware AI: Agentic Risk Auditing of Soft Prompt Attacks on ESM-Based Variant Predictors
-
Deep Ignorance: Filtering Pretraining Data Builds Tamper-Resistant Safeguards into Open-Weight LLMs
-
Deep Research Brings Deeper Harm
-
Exposing Critical Safety Failures: A Comprehensive Safety-Weighted Evaluation of LLaMA Models for Biochemical Toxicity Screening
-
GeneBreaker: Jailbreak Attacks against DNA Language Models with Pathogenicity Guidance
-
Involuntary Jailbreak
-
Is My Language Model a Biohazard?
-
Monte Carlo Expected Threat (MOCET) Scoring
-
Open-weight genome language model safeguards: Assessing robustness via adversarial fine-tuning
-
Perspective: Lessons from Cybersecurity for Biological AI Safety
-
Position: Biosafety-Critical Adjacent Technologies are Critical for Scalable and Safe Clinical Multi-modal LLM Deployment
-
Position: Without Global Governance, AI-Enabled Biodesign Tools Risk Dangerous Proliferation
-
Prompting Toxicity: Analyzing Biosafety Risks in Genomic Language Models
-
Property Adherent Molecular Generation with Constrained Discrete Diffusion
-
ProtGPT2 is Not Biosecure by Default
-
Resisting RL Elicitation of Biosecurity Capabilities: Reasoning Models Exploration Hacking on WMDP
-
RippleBench: Capturing Ripple Effects by Leveraging Existing Knowledge Repositories
-
Robust LLM Unlearning with MUDMAN: Meta-Unlearning with Disruption Masking And Normalization
-
SafeBench-Seq: A Homology-Clustered, CPU-Only Baseline for Protein Hazard Screening with Physicochemical/Composition Features and Cluster-Aware Confidence Intervals
-
SafeGenie: Erasing Dangerous Concepts from Biological Diffusion Models
-
SafeProtein: Red-Teaming Framework and Benchmark for Protein Foundation Models
-
Securing Dual-Use Pathogen Data of Concern
-
Securing the Language of Life: Inheritable Watermarks from DNA Language Models to Proteins
-
Structural Persistence Despite Sequence Redaction: A Biosecurity Evaluation of Protein Language Models
-
Translating Biomedical Observations into Signal Temporal Logic with LLMs using Structured Feedback
-
Where to Edit? : Complementary Protein Property Control from Weight and Activation Spaces
-
Without Safeguards, AI-Biology Integration Risks Creating Future Pandemics
-
Zero-Shot Embedding Drift Detection: A Lightweight Defense Against Prompt Injections in LLMs