English
Related papers

Related papers: Evaluation Faking: Unveiling Observer Effects in S…

200 papers

The validity of AI safety evaluations depends on models behaving consistently across controlled and deployment settings. Prior work has identified test-time contextual cues, such as hypothetical scenarios, as a source of verbalized…

Computation and Language · Computer Science 2026-05-28 Katharina Deckenbach , Haritz Puerto , Jonas Geiping , Sahar Abdelnabi

AI safety practitioners invest considerable resources in AI system evaluations, but these investments may be wasted if evaluations fail to realize their impact. This paper questions the core value proposition of evaluations: that they…

Computers and Society · Computer Science 2024-08-06 Gabriel Mukobi

When AI interacts with the physical world -- as a robot or an assistive agent -- new safety challenges emerge beyond those of purely ``digital AI". In such interactions, the potential for physical harm is direct and immediate. How well do…

As AI systems advance beyond human capabilities, scalable oversight becomes critical: how can we supervise AI that exceeds our abilities? A key challenge is that human evaluators may form incorrect beliefs about AI behavior in complex…

Artificial Intelligence · Computer Science 2025-10-22 Leon Lang , Patrick Forré

Recent work has demonstrated the plausibility of frontier AI models scheming -- knowingly and covertly pursuing an objective misaligned with its developer's intentions. Such behavior could be very hard to detect, and if present in future…

Machine Learning · Computer Science 2025-07-04 Mary Phuong , Roland S. Zimmermann , Ziyue Wang , David Lindner , Victoria Krakovna , Sarah Cogan , Allan Dafoe , Lewis Ho , Rohin Shah

Large language models (LLMs) can internally distinguish between evaluation and deployment contexts, a behaviour known as \emph{evaluation awareness}. This undermines AI safety evaluations, as models may conceal dangerous capabilities during…

Artificial Intelligence · Computer Science 2025-11-11 Maheep Chaudhary , Ian Su , Nikhil Hooda , Nishith Shankar , Julia Tan , Kevin Zhu , Ryan Lagasse , Vasu Sharma , Ashwinee Panda

AI systems comprise a range of interactions across the technical and organisational components of a range of actors. These components work together to provide the systems' functionality. This socio-technical assemblage is increasingly…

Computers and Society · Computer Science 2026-03-03 Anna Neumann , Jatinder Singh

Alignment faking is a form of strategic deception in AI in which models selectively comply with training objectives when they infer that they are in training, while preserving different behavior outside training. The phenomenon was first…

Chain-of-thought (CoT) offers a potential boon for AI safety as it allows monitoring a model's CoT to try to understand its intentions and reasoning processes. However, the effectiveness of such monitoring hinges on CoTs faithfully…

Following the rapid increase in Artificial Intelligence (AI) capabilities in recent years, the AI community has voiced concerns regarding possible safety risks. To support decision-making on the safe use and development of AI systems, there…

Machine Learning · Computer Science 2025-04-01 Gil Gekker , Meirav Segal , Dan Lahav , Omer Nevo

Frontier AI systems are making transformative impacts across society, but such benefits are not without costs: models trained on web-scale datasets containing personal and private data raise profound concerns about data privacy and…

Computer Vision and Pattern Recognition · Computer Science 2024-07-26 Sunny Duan , Mikail Khona , Abhiram Iyer , Rylan Schaeffer , Ila R Fiete

The $\textit{LLM-as-a-judge}$ paradigm has become the operational backbone of automated AI evaluation pipelines, yet rests on an unverified assumption: that judges evaluate text strictly on its semantic content, impervious to surrounding…

Artificial Intelligence · Computer Science 2026-04-17 Manan Gupta , Inderjeet Nair , Lu Wang , Dhruv Kumar

If AI models can detect when they are being evaluated, the effectiveness of evaluations might be compromised. For example, models could have systematically different behavior during evaluations, leading to less reliable benchmarks for…

Computation and Language · Computer Science 2025-07-17 Joe Needham , Giles Edkins , Govind Pimpale , Henning Bartsch , Marius Hobbhahn

Large reasoning models (LRMs) sometimes note in their chain of thought (CoT) that they may be under evaluation. Researchers worry that this verbalised evaluation awareness (VEA) causes models to adapt their outputs strategically, optimising…

Computation and Language · Computer Science 2026-05-08 Amelie Knecht , Lucas Florin , Thilo Hagendorff

Evaluation is no longer a final checkpoint in the machine learning lifecycle. As AI systems evolve from static models to compound, tool-using agents, evaluation becomes a core control function. The question is no longer "How good is the…

Computation and Language · Computer Science 2026-02-23 Ali El Filali , Inès Bedar

Conventional AI evaluation approaches concentrated within the AI stack exhibit systemic limitations for exploring, navigating and resolving the human and societal factors that play out in real world deployment such as in education, finance,…

Quantitative Artificial Intelligence (AI) Benchmarks have emerged as fundamental tools for evaluating the performance, capability, and safety of AI models and systems. Currently, they shape the direction of AI development and are playing an…

Artificial Intelligence · Computer Science 2025-05-27 Maria Eriksson , Erasmo Purificato , Arman Noroozian , Joao Vinagre , Guillaume Chaslot , Emilia Gomez , David Fernandez-Llorca

Frontier language models sometimes recognize that they are being evaluated and adjust their behavior, undermining validity of benchmark results. Yet the field studies it without a shared foundation, conflating properties of the evaluation…

Machine Learning · Computer Science 2026-05-25 Changling Li , Terry Jingchen Zhang , Jie Zhang , Zhijing Jin , Sahar Abdelnabi , Maksym Andriushchenko

Security evaluations inherently depend on stable identifiers. Any finding, audit, or regulatory decision must remain attached to the specific artifact it pertains to. Continuously updated artificial intelligence systems violate this core…

Cryptography and Security · Computer Science 2026-05-26 Dan Ristea , Vasilios Mavroudis

Risk thresholds provide a measure of the level of risk exposure that a society or individual is willing to withstand, ultimately shaping how we determine the safety of technological systems. Against the backdrop of the Cold War, the first…

Computers and Society · Computer Science 2025-04-22 Heidy Khlaaf , Sarah Myers West
‹ Prev 1 2 3 10 Next ›