The paper introduces MIRAGE, a benchmark of 750 multi‑step decision tasks designed to test autonomous web agents’ investigative abilities across Wikipedia Forensics, Shopping Admin adjudication, and Reddit Moderation. Each task contains a misleading visible context and a hidden context that holds decisive evidence, allowing performance to be broken down into Investigation, Reasoning, and Decision Accuracy, with an added Investigative Hallucination Rate. Evaluation of eight LLM agents reveals three consistent patterns: agents often reach relevant pages but fail to extract decisive evidence, procedural hints improve investigation but not decision accuracy on Wikipedia tasks, and 12.6% of trajectories include fabricated facts.
arXiv:2607. 12267v1 Announce Type: cross Abstract: Language agents that interleave reasoning and tool use degrade sharply as reasoning chains lengthen, even when each individual step is easy.
arXiv:2607. 17291v1 Announce Type: new Abstract: Deep research agents increasingly operate over the open web, where relevant records coexist with redundant summaries, outdated reports, and misleading documents.
arXiv:2609.06027v1 Announce Type: cross Abstract: Search-augmented LLM agents are increasingly used for consumer decisions, making them vulnerable to Generative Engine Optimization (GEO) poisoning. E...
arXiv:2608. 02011v2 Announce Type: replace Abstract: Agentic retrieval-augmented generation (RAG) systems can fail before evidence-conditioned reasoning is tested: an agent may retrieve candidate snippets but finalize without inspecting them.
arXiv:2606. 26071v1 Announce Type: new Abstract: A central goal of safety research is determining whether a model is misaligned.
arXiv:2606. 08671v1 Announce Type: new Abstract: Agent skills extend language-model agents with task-specific procedures, scripts, and references, but the tasks and environments they target continually change.
arXiv:2605. 12887v2 Announce Type: replace-cross Abstract: Web-enabled LLM agents are changing how online information influences search outcomes.
Platform abuse campaigns conceal redirection instructions with emojis, homophones, character decomposition, and redundant symbols, then route users through disguised links to services associated with...
arXiv:2606. 02060v1 Announce Type: new Abstract: Deep-research agents solve tasks through long trajectories of search, tool use, evidence inspection, and answer synthesis.
arXiv:2608. 12877v1 Announce Type: new Abstract: Multi-hop fact verification, which verifies claims by reasoning over multiple pieces of evidence, is critical for combating misinformation on social media yet remains highly challenging.
arXiv:2606. 22737v2 Announce Type: replace Abstract: Before letting an agent operate over real context, can you prove it used the right evidence?
arXiv:2606. 00644v1 Announce Type: new Abstract: AI research often requires decisions before future evidence exists: which bottleneck to attack, which direction to pursue, or where a project should be positioned.