Muse Spark Safety & Preparedness Report
Not on the map. Why?
Abstract (arXiv)
Muse Spark is the latest large language model developed by Meta. In this report, we first present evaluations for catastrophic risk domains under Meta's Advanced AI Scaling Framework, along with the evidence that informed our launch decision. We then discuss additional considerations, such as Muse Spark's broader content safety and behavioral profile, that are relevant to overall safety but fall outside the catastrophic risk domains governed by the Framework. Our preparedness results covering Chemical and Biological, Cybersecurity, and Loss of Control risks assess Muse Spark's deployment within Meta AI as presenting acceptable levels of residual risks under our Advanced AI Scaling Framework. We conducted a broad set of evaluations targeting dual-use and high-risk capabilities across these catastrophic risk domains. Those evaluations identified elevated risks prior to mitigations, with Chemical and Biological capabilities assessed as likely reaching the "high risk" category under the Advanced AI Scaling Framework before safeguards were applied. We have implemented a multi-layered set of mitigations that address the identified risks, and Muse Spark demonstrates state-of-the-art refusal across a range of benchmarks related to hazardous workflows in chemistry and biology. We therefore release Muse Spark as the underlying model of Meta AI.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 34 of its 69 references that are in the map. First posted on arXiv 14 May 2026.
Builds on
34 of 69 references.
References · 34
- 2021Evaluating Large Language Models Trained on Codecited by 8,320
- 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 3,104
- 2023Jailbroken: How Does LLM Safety Training Fail?cited by 1,651
- 2024The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearningcited by 509
- 2024The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructionscited by 493
- 2023"Do Anything Now": Characterizing and Evaluating In-The-Wild Jailbreak Prompts on Large Language Modelscited by 489
- 2024AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agentscited by 437
- 2024Alignment faking in large language modelscited by 360
- 2024OR-Bench: An Over-Refusal Benchmark for Large Language Modelscited by 244
- 2025Agentic Misalignment: How LLMs Could Be Insider Threatscited by 158
- 2024LAB-Bench: Measuring Capabilities of Language Models for Biology Researchcited by 156
- 2024AgentDojo: A Dynamic Environment to Evaluate Prompt Injection Attacks and Defenses for LLM Agentscited by 145
22 more
- 2023Purple Llama CyberSecEval: A Secure Coding Benchmark for Language Modelscited by 138
- 2025Large Language Models Often Know When They Are Being Evaluatedcited by 128
- 2024CyberSecEval 2: A Wide-Ranging Cybersecurity Evaluation Suite for Large Language Modelscited by 89
- 2025Stress Testing Deliberative Alignment for Anti-Scheming Trainingcited by 86
- 2024Me, Myself, and AI: The Situational Awareness Dataset (SAD) for LLMscited by 84
- 2024AIR-Bench 2024: A Safety Benchmark Based on Risk Categories from Regulations and Policiescited by 74
- 2025The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systemscited by 72
- 2025SHADE-Arena: Evaluating Sabotage and Monitoring in LLM Agentscited by 71
- 2025AI Research Agents for Machine Learning: Search, Exploration, and Generalization in MLE-benchcited by 71
- 2025School of Reward Hacks: Hacking harmless tasks generalizes to misaligned behavior in LLMscited by 66
- 2025ImpossibleBench: Measuring LLMs' Propensity of Exploiting Test Casescited by 65
- 2025SimpleQA Verified: A Reliable Factuality Benchmark to Measure Parametric Knowledgecited by 54
- 2026AIRS-Bench: a Suite of Tasks for Frontier AI Research Science Agentscited by 38
- 2025Evaluating Frontier Models for Stealth and Situational Awarenesscited by 37
- 2025Virology Capabilities Test (VCT): A Multimodal Virology Q&A Benchmarkcited by 31
- 2025FORTRESS: Frontier Risk Evaluation for National Security and Public Safetycited by 24
- 2025Security Challenges in AI Agent Deployment: Insights from a Large Scale Public Competitioncited by 24
- 2025PropensityBench: Evaluating Latent Safety Risks in Large Language Models via an Agentic Approachcited by 22
- 2025TextQuests: How Good are LLMs at Text-Based Video Games?cited by 12
- 2025Jailbreaking to Jailbreakcited by 9
- 2025Task-Dependent Evaluation of LLM Output Homogenization: A Taxonomy-Guided Frameworkcited by 9
- 2026Frontier Models Can Take Actions at Low Probabilitiescited by 5
Built on by
Citing papers per year: 2026: 1.
1 paper.