{"slug": "agentrisk-automation-is-not-intent-measuring-how-fraud-models-fail-legitimate-ai", "title": "AgentRisk: automation is not intent, measuring how fraud models fail legitimate AI shopping agents", "summary": "Rishabh released AgentRisk, a simulated benchmark and preprint measuring how fraud models falsely decline legitimate AI shopping agents, with a demo Space hosted in an AgentRisk collection. The benchmark reports that a verified-agent approach, AgentGate, fixes the verified-agent problem but drops hijack stopping from 96.1% to 70.2%, illustrating the tradeoff between failure modes. Rishabh states the work is not peer-reviewed and that the public data is \"a starting point for discussion and replication, not a verdict on any vendor.", "body_md": "Hi everyone, I am Rishabh. Here is a problem I keep coming back to: AI shopping agents now buy things on behalf of real people, and fraud models keep flagging them. To a fraud model, an agent looks like a bot: fast, systematic, no human hesitation. But automation is not intent. A legitimate agent completing a real purchase is not fraud, and every false decline is a real customer losing a real order.\n\nAgentRisk is my attempt to measure that failure properly.\n\nWhat it is:\n\nThe tradeoff is real and it is the interesting part: AgentGate fixes the verified-agent problem, but hijack stopping drops from 96.1% to 70.2%. Fixing one failure mode costs you on another, and that tension is exactly what the benchmark is for.\n\nHonest scope: this is a simulated benchmark and a preprint, not peer-reviewed work, and the public data is a starting point for discussion and replication, not a verdict on any vendor.\n\nTry it:\n\nThe demo Space sits in the AgentRisk collection on my profile. If you work on agent safety, fraud, or evals, I would love your take: where does this failure mode show up in your systems, and what would you add to the benchmark? The writeup is in progress, and feedback here will shape it.", "url": "https://wpnews.pro/news/agentrisk-automation-is-not-intent-measuring-how-fraud-models-fail-legitimate-ai", "canonical_source": "https://discuss.huggingface.co/t/agentrisk-automation-is-not-intent-measuring-how-fraud-models-fail-legitimate-ai-shopping-agents/191160#post_1", "published_at": "2026-10-28 20:39:07+00:00", "updated_at": "2026-10-28 20:44:49.727361+00:00", "lang": "en", "topics": ["ai-agents", "ai-safety", "ai-research"], "entities": ["Rishabh", "AgentRisk", "AgentGate"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/agentrisk-automation-is-not-intent-measuring-how-fraud-models-fail-legitimate-ai", "markdown": "https://wpnews.pro/news/agentrisk-automation-is-not-intent-measuring-how-fraud-models-fail-legitimate-ai.md", "text": "https://wpnews.pro/news/agentrisk-automation-is-not-intent-measuring-how-fraud-models-fail-legitimate-ai.txt", "jsonld": "https://wpnews.pro/news/agentrisk-automation-is-not-intent-measuring-how-fraud-models-fail-legitimate-ai.jsonld"}}