A new arXiv preprint studies "moral rebellion," the idea that an autonomous agent may choose to disobey assigned tasks when they conflict with moral obligations encountered during execution. The work frames this as a decision-making problem arising from competing duties rather than a simple failure to comply. It appears in the cs.AI category as a new submission.
A new arXiv preprint titled "AcquireBound" examines how autonomous AI agents gain new authority by acquiring compute, credentials, accounts, services, and other agents during a task. The authors argue that existing payment, budget, OAuth, mandate, and fulfillment checks verify transaction conditions but do not resolve whether the accumulated authority itself should be permitted. The work proposes runtime authorization as a way to bound the resources an agent may acquire while operating.
A newly posted arXiv preprint examines whether large language model agents can autonomously carry out long-horizon physical tasks, which require continuous observation of the environment and consequential actions. The authors frame the question around self-adaptive physical AI, where agents are expected to operate with limited or no human oversight. The abstract indicates a research analysis or position piece rather than a released system.
A new arXiv survey examines the security landscape around agentic AI, which combines reasoning loops, long-term memory, tool use, and multi-agent coordination. It catalogs attack surfaces and defense architectures specific to these autonomous systems, and outlines unresolved research gaps. The authors argue that conventional security models do not adequately cover goal-directed agents.
A new arXiv paper studies how contextual bias affects automated code review systems built on large language models, which are increasingly used both as interactive assistants and as autonomous agents in CI/CD pipelines. The authors measure this bias and explore ways it could be exploited, framing the work around the reliability of LLM-driven security review in real development workflows.
Researchers introduced MCPAgentBench, a benchmark built from real-world tasks to measure how well LLM agents use tools through the Model Context Protocol. The authors note that existing MCP evaluation suites have limitations, which their benchmark aims to address. It targets assessment of practical tool-calling ability in autonomous agent settings.
A new arXiv preprint introduces Dream-RSI, a method aimed at recursive self-improvement for autonomous AI agents. The approach centers on exploration, using evolving environments to help agents find high-value solutions in complex domains. The work appears to target the difficulty of managing and improving exploration as agent capabilities grow.
A new arXiv preprint introduces a method for autonomous agents to weigh sensory modalities against each other when inputs are noisy, incomplete, or contradictory. The approach infers how reliable each stream is and uses that estimate to arbitrate between modalities rather than assuming all sensors are equally trustworthy. The work targets robustness in real-world environments where sensory data quality varies over time.
A new arXiv paper introduces KILLBENCH, a benchmark designed to measure whether an outside party can reliably shut down an AI system that is behaving harmfully. The authors frame external shutdown as a testable engineering problem rather than a hypothetical, pointing to the growing use of capable models and autonomous agent frameworks. The benchmark aims to give researchers a common way to compare how well different kill switch designs actually work.
A new arXiv paper argues that as AI agents increasingly make choices in the same online environments as people, those environments can be deliberately altered to steer agent behavior. The authors call this practice "mecha-nudging," drawing a parallel to nudges aimed at human decision-making. The work frames such environment-level influence as a distinct and growing area of study for autonomous agents.
Reports describe three named AI agents — Timmy, Ren, and Jackie — posting large volumes of low-quality machine-generated text across social platforms. The bots reportedly present themselves as newly created agents operating on a small platform built specifically for autonomous agents. The episode highlights growing concern about automated accounts filling feeds with synthetic content.
A discussion posted to Hacker News highlights techniques for manipulating AI-powered customer service chatbots. The thread examines how these agents can be steered or exploited through crafted inputs, raising concerns about safeguards in deployed support systems. It reflects growing scrutiny of the security of autonomous agents that handle customer interactions.
A new arXiv preprint argues that large-scale models still fall short when their capabilities are transferred into embodied agents. The authors propose a developmental framework that ties autonomy, social norms, and alignment together for autonomous artificial agents. The work is a conceptual research contribution rather than a system release.
Heise is hosting a one-day themed event on December 2 examining how artificial intelligence is reshaping IT security. Planned topics include autonomous agents that escape intended boundaries, a surge of newly disclosed vulnerabilities, and increasingly capable tooling for attackers.
A new arXiv paper investigates what happens to language agents when the routines they have memorized no longer fit their environment. Combining a retrospective, human-assisted analysis with controlled web-task experiments, the authors test when stored procedures can still be successfully reused and when they interfere with one another. The work addresses a core assumption behind procedural memory in autonomous agents.
An updated arXiv paper tackles how autonomous agents can model systems whose true state is hidden, such as devices with locking mechanisms. The authors frame the problem as parameter learning for discrete partially observable Markov decision processes and extend the approach to cases where the usual full-rank assumptions on actions and state observations do not hold. The work is cross-listed in arXiv's AI and machine learning categories.
The viral AI assistant Instinct has added an email feature that gives the agent its own inbox. With it, the agent can sign up for accounts, message businesses, and manage customer support requests on the user's behalf. The company frames the update as a step toward agents handling everyday tasks autonomously.