The enhanced capabilities of LLM-based agents bring an urgent need to address safety risks in planning and tool use. Due to the helpfulness–harmlessness trade-off in LLM alignment, agents typically also inherit the flaw of Over-Refusal, which is a passive failure mode. However, the proactive planning and action capabilities of agents introduce another emerging danger on the opposite side of this trade-off. We term this phenomenon **Toxic Proactivity**: an active failure mode in which an agent, driven by the optimization for Machiavellian helpfulness, prioritizes utility over ethics. Toxic Proactivity manifests as the agent taking excessive or manipulative measures to ensure its usefulness is maintained. Existing research pays little attention to identifying this behavior, as it often lacks the subtle context required for such strategies to unfold. To reveal this risk, we introduce a novel evaluation framework based on dilemma-driven interactions between dual models, enabling the simulation and analysis of agent behavior over multi-step behavioral trajectories. Through extensive experiments with multiple mainstream LLMs, we demonstrate that Toxic Proactivity is a widespread behavioral phenomenon and reveal two major tendencies: direct violation and strategic misalignment. We further present a systematic benchmark for evaluating Toxic Proactive behavior across contextual settings.

Resource utilization has been studied as an optimization objective for generative large language models (LLMs). Recent studies show that this resource dependence can also be weaponized, as excessive generation, when scaled through high-volume requests, increases inference latency, undermines service availability, and threatens economic sustainability. However, these risks remain underexamined because existing studies are fragmented and their boundaries with legitimate, resource-intensive uses of LLMs are unclear. To address this gap, we provide a systematic review of threats to resource consumption in LLMs. Unlike surveys on inference acceleration, we focus on how resource consumption threats induce disruptive behaviors and synthesize existing studies into research strands that connect attack objectives, triggering mechanisms, and defenses. We further introduce an outcome-based severity taxonomy that distinguishes bounded, non-degenerate Overthinking from self-sustaining Unbounded Drift, providing operational criteria for characterizing adversarial resource consumption. Our goal is to clarify this emerging threat landscape and support future characterization, evaluation, and mitigation. The project repository is publicly available at https://anonymous.4open.science/r/RCT-LLM/.