Related papers: SpecEval: Evaluating Model Adherence to Behavior S…
Reports of human-like behaviors in foundation models are growing, with psychological theories providing enduring tools to investigate these behaviors. However, current research tends to directly apply these human-oriented tools without…
Frontier AI developers now train models against long written behavioral specifications, such as Anthropic's constitution (Anthropic, 2025a) and OpenAI's Model Spec (OpenAI, 2025a), integrated into post-training via methods like character…
Foundation models are critical digital technologies with sweeping societal impact that necessitates transparency. To codify how foundation model developers should provide transparency about the development and deployment of their models, we…
The rapid deployment of AI agents in commercial settings has outpaced the development of evaluation methodologies that reflect production realities. Existing benchmarks measure agent capabilities through retrospectively curated tasks with…
As autonomous AI agents are increasingly deployed in high-stakes environments, ensuring their safety and alignment with human values is becoming a practical deployment concern. Current benchmarks for AI agents primarily evaluate refusal of…
Following the AI Seoul Summit in 2024, twelve AI companies published frontier AI safety frameworks (Frameworks) outlining their approaches to managing catastrophic risks from advanced AI systems. Emerging legislation increasingly treats…
As foundation models grow in both popularity and capability, researchers have uncovered a variety of ways that the models can pose a risk to the model's owner, user, or others. Despite the efforts of measuring these risks via benchmarks and…
Reliable pattern recognition systems should exhibit consistent behavior across similar inputs, and their explanations should remain stable. However, most Explainable AI evaluations remain instance centric and do not explicitly quantify…
Model cards are the primary documentation framework for developers of artificial intelligence (AI) models to communicate critical information to their users. Those users are often developers themselves looking for relevant documentation to…
Foundation models have rapidly permeated society, catalyzing a wave of generative AI applications spanning enterprise and consumer-facing contexts. While the societal impact of foundation models is growing, transparency is on the decline,…
Foundation models are routinely fine-tuned for use in particular domains, yet safety assessments are typically conducted only on base models, implicitly assuming that safety properties persist through downstream adaptation. We test this…
Recent advancements in foundation models have improved autonomous tool usage and reasoning, but their capabilities in map-based reasoning remain underexplored. To address this, we introduce MapEval, a benchmark designed to assess foundation…
The range of application of artificial intelligence (AI) is vast, as is the potential for harm. Growing awareness of potential risks from AI systems has spurred action to address those risks, while eroding confidence in AI systems and the…
As foundation models have accumulated hundreds of millions of users, developers have begun to take steps to prevent harmful types of uses. One salient intervention that foundation model developers adopt is acceptable use policies: legally…
Large language models (LLMs) are increasingly trained from AI constitutions and model specifications that establish behavioral guidelines and ethical principles. However, these specifications face critical challenges, including internal…
In this position paper, we argue that human baselines in foundation model evaluations must be more rigorous and more transparent to enable meaningful comparisons of human vs. AI performance, and we provide recommendations and a reporting…
Evaluating alignment in language models requires testing how they behave under realistic pressure, not just what they claim they would do. While alignment failures increasingly cause real-world harm, comprehensive evaluation frameworks with…
We introduce a framework for Foundational Analysis of Safety Engineering Requirements (SAFER), a model-driven methodology supported by Generative AI to improve the generation and analysis of safety requirements for complex safety-critical…
Foundation models - models trained on broad data that can be adapted to a wide range of downstream tasks - can pose significant risks, ranging from intimate image abuse, cyberattacks, to bioterrorism. To reduce these risks, policymakers are…
Despite over 3.5 million Android apps and 200+ million Android Auto-compatible vehicles, only a few hundred apps support Android Auto due to platform-specific compliance requirements. Android Auto mandates service-based architectures in…