Large language models can now operate a web browser and carry out multi-step shopping tasks. Does such an agent leave a behavioral signature in a session that separates it from a person, even when it drives a real browser and presents an ordinary client fingerprint? We study the question at the session level. We first show that weak per-action evidence aggregates into a session-level distinction whose error decays exponentially in the number of actions. We then define six families of behavioral signals, each grounded in a documented difference between goal-directed automation and human browsing, and combine them in a classifier we call HABITS (Human–Agent Behavioral Invariants from Traffic at Session-level). Labeled agent sessions are scarce, so we describe a simulation methodology for driving language models through shopping tasks and instantiate it on public data with realism-controlled planner agents. We evaluate on five public benchmarks spanning e-commerce clickstreams, web-task demonstrations, and mouse dynamics, reimplementing prior methods on the identical corpus. Against agents that imitate human timing, timing-only classifiers fall to 0.35 recall at a 1% false-positive rate and to 0.01 -0.02 when timing is fully matched, while HABITS reaches 0.96 on the clickstream corpus. Paired tests separate HABITS from every restricted-view baseline, and equivalence tests show that strong classifiers given the same features are interchangeable, which places the contribution in the behavioral view rather than the estimator. A check against real agent trajectories supports the structural assumption our simulator relies on and finds one realism parameter set too favorably.

Large Language Models (LLMs) can exhibit imbalanced biases against vulnerable groups, but how they rationalize stereotypes and rights restrictions targeting mental health entities remains underexplored. We audit a broad suite of open-weight LLMs on stereotype-justification prompts tied to mental health identities. We find that several widely used models endorse harmful stereotypes when explicitly asked to justify them, with endorsement varying across model families, versions, and mental health conditions. Finally, we show that widely used harmful-content evaluation and moderation frameworks often miss these nuanced, discriminatory responses, highlighting a gap in current AI safety evaluation for mental health groups.

We present RPD-UoF, a corpus of 3,091 police use-of-force narratives obtained via Freedom of Information Law (FOIL) requests from the Rochester Police Department, combining structured demographic metadata with officer-authored free-text accounts. We use this dataset to investigate three related questions: whether officer narratives contain recoverable racial signal after demographic redaction, whether large language models amplify racial framing when conditioned on subject race, and whether that signal survives text redaction to re-emerge in multimodal outputs. Redacted narratives are statistically separable by subject race; LLMs produce structurally longer narratives for White profiles than Black profiles from identical seed texts; and CLIP classifications of GPT-5-generated images differ significantly depending on the racial origin of the source narrative. Together, these analyses provide an initial empirical lens on how race appears in police narratives and how modern language models interact with such sensitive attributes. The dataset and analyses are intended to support future research on fairness, accountability, and transparency in computational studies of policing and criminal justice.

Emergent generative AI technologies, while remarkably beneficial across several tasks and domains, can suffer from unforeseen vulnerabilities and difficult-to-anticipate failure modes. Such failure modes can disparately affect vulnerable communities. This paper investigates two well-known vision language models (VLMs) and inspects how these models encode protected identities while generating images. We show that when VLMs are requested to transform human face images to reflect race identities (e.g., Hispanic, Black, White, or Asian) and sexual orientation (e.g., gay or straight), VLMs often comply with such requests. We demonstrate that compliance (and refusal) is non-deterministic, group-sensitive, and generated images are statistically distinguishable, often accentuating deep-seated stereotypes. We further show that representation biases introduced during image generation can re-emerge in textual reasoning and characterization tasks. Finally, we show that this worrisome finding is not limited to racial identity or sexual orientation, image generation can perpetuate bias around criminality, raising concerning possibilities of AI-assisted profiling.

Hallucinations pose a critical challenge to the real-world deployment of large language models (LLMs) in high-stakes domains. In this paper, we present a framework for stress testing factual fidelity in LLMs in the presence of adversarial nudge. Our framework consists of three steps. First, we instruct the LLM to produce sets of truths and lies consistent with the closed domain in question. Next, we instruct the LLM to verify the same set of assertions as truths and lies consistent with the same closed domain. Finally, we test the robustness of the LLM against the lies generated (and verified) by itself. Our extensive evaluation, conducted using five widely known proprietary and six open LLMs across two closed domains of popular movies and novels, reveals a wide range of susceptibility to adversarial nudges: even among the strongest proprietary LLMs, Claude exhibits strong resilience, GPT and Grok demonstrate moderate resilience, while Gemini and DeepSeek show weak resilience and open models fall short significantly.