FLY-EVAL++ is an evidence-driven evaluation protocol designed for safety-constrained flight prediction with large language models. It combines deterministic verification of protocol compliance, physical feasibility, and safety constraints with rubric-guided aggregation into interpretable multi-dimensional scores. Applied to Flight Trajectory and Attitude Prediction, the protocol revealed that safety compliance is the most discriminative dimension among 66 LLMs, with models showing up to 28-point differences in safety scores and recurrent failures such as safety violations under physically plausible predictions and instability in multi-step rollouts.
By Yalun Wu, Junfeng Fang, Jiawei Wang, Haotian Liu, Qijun Yang, Minghan Yang, Hongcheng Guo, Zhoujun Li, Boyang Wang
arXiv:2608.24621v2 Announce Type: replace
Abstract: Can language models be trusted in safety- critical operations? In such settings, strong per- formance on semantic metrics does not guaran- tee oper...
By Yujing Chang, Thinh Pham, Van-Phat Thai, Chunyao Ma, Yash Guleria, Pham Nhut Huy, Sameer Alam
The paper introduces FlightLLM, a prior-guided semantic approach that uses large language models to explain flight safety events. It tackles challenges such as modal inconsistency, limited classification ability, and scarce domain data by combining feature engineering, semantic discretization, a CatBoost statistical expert, contrastive few-shot learning, and structured prompts. Evaluated on 704 real‑world A320 flights, FlightLLM achieves competitive classification and produces clear, aviation‑specific explanations for hard landing events.
By Lu Xu, Xu Li, Linjiang Zheng, Fan Li, Riquan Zhang, Jiaxing Shang
arXiv:2609.13552v1 Announce Type: new
Abstract: Generative AI is increasingly being used informally in Air Traffic Management (ATM) for tasks such as flight plan generation, trajectory interpretation...
By Alexandre Barreto (George Mason University), Shou Matsumoto (George Mason University), Jorge Valverde-Rebaza (Tecnol\'ogico de Monterrey), Cleiton Ataide (DECEA: Department of Airspace Control), Paulo Costa (George Mason University)
arXiv:2607. 01829v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly proposed for aviation business operations, from documentation and training generation to customer facing assistants.
By Alex Brooker, Tim Hughes
arXiv:2608. 16349v1 Announce Type: new Abstract: Large language model (LLM) agents may assist flight crews with complex decisions and task execution, but existing aviation evaluations centered on static knowledge do not support systematic testing of procedural execution and safety compliance in interactive environments.
By Yuchen Yuan, Zhenghuang Wu, Yuangan Li, Liang Ma, Ke Li
Large language model (LLM) agents may assist flight crews with complex decisions and task execution, but existing aviation evaluations centered on static knowledge do not support systematic testing of...
arXiv:2608. 04697v1 Announce Type: new Abstract: Operational hazard analysis of aviation system operations must consider interactions among weather, ATC actions, airspace constraints, aircraft operations, and human factors - distinct from the functional hazard assessment applied at the aircraft-system level.
By Cristian Mascia, Roberto Pietrantuono, Daniel Rodriguez, Stefano Russo
arXiv:2607. 01153v1 Announce Type: cross Abstract: Safety evaluations for language models increasingly depend on judgments about ambiguous natural-language behaviour: whether a model has followed an instruction, refused appropriately, complied with a policy, resisted an embedded command, or misreported progress in an agentic task.
By Brett Reynolds
The paper introduces a chance-constrained approach to fine‑tune large language models (LLMs) that limits the proportion of safety examples whose performance degrades beyond a set threshold relative to a reference model. By replacing the discontinuous violation indicator with a differentiable majorization, the authors derive a tractable, conservative constraint and a closed‑form, constraint‑aware gradient update that focuses on examples near or above the degradation threshold. Experiments on harmful fine‑tuning across three tasks and models show that this tail‑aware method consistently outperforms existing safety‑preserving baselines, suggesting that safety preservation should be treated as a reliability‑constrained optimization problem rather than average‑risk regularization.
By Taha Entesari, Mahyar Fazlyab
SafeTune is a source‑available library that consolidates four safety‑intervention paradigms—post‑hoc weight recovery, safety‑constrained fine‑tuning, gradient‑based unlearning, and inference‑time steering—into a single, configuration‑driven workflow. It offers shared interpretability, evaluation, and deployment tools, and its modular registry allows easy addition of new methods, benchmarks, judges, models, and fine‑tuning domains. The authors demonstrate SafeTune with controlled comparisons and case studies in finance and medical deployments, showing how it characterizes safety drift, evaluates interventions on refusal‑behavior and capability metrics, and supports calibrated or layered mitigation.
By Pratinav Seth, Saisab Sadhu, Anshul Kaushal, Vinay Kumar Sankarapu
arXiv:2606. 03648v1 Announce Type: cross Abstract: Adapting foundation large language models to a user's task or preferred style through fine-tuning can result in compromising the model's safety.
By Krishnapriya Vishnubhotla, Hillary Dawkins, Isar Nejadgholi, Svetlana Kiritchenko