FEND: Far-End Detection in Outbound Voice AI Calls A proposed framework called FEND detects what answers an outbound Voice AI call — a live person, voicemail, an IVR system or another AI — and returns a single label within 20 seconds, according to an arXiv paper (2611.18050v1). In 1,000 randomly sampled calls from the authors' own outbound deployment, a live person answered 48.1%, a voicemail system 44.6%, an AI 6.9% and an IVR 0.4%; standard Answering Machine Detection reports only human versus machine, which cannot distinguish those automated classes. On production traffic, separating a live person from any automated far end reached F1 0.979, with only 2.9% of live people mislabeled as automated, and speech timing separated AI from human at 0.952 ROC-AUC because synthesized speech leaves far shorter gaps between words. arXiv:2611.18050v1 Announce Type: new Abstract: When a conversational Voice AI agent places an outbound call, the far end may be a live person, a voicemail greeting, an interactive voice response IVR system, or increasingly another AI system answering on the callee's behalf. Standard Answering Machine Detection AMD reports only whether a human or a machine picked up, a design meant for human call centres, where one machine is as uninteresting as another. That single bit does not tell a Voice AI agent what to do, since leaving a voicemail message, navigating an IVR and holding a short exchange with another AI are three different actions and a two-valued label cannot distinguish them. In 1,000 randomly sampled calls from our own outbound deployment, a live person answered 48.1%, a voicemail system 44.6%, an AI 6.9% and an IVR 0.4%. Our proposed framework, FEND, reads the far end's audio and its streaming transcript and runs one detector for each automated class voicemail, IVR, AI callee , with human the default when nothing fires, returning a single label within 20 seconds. On a public benchmark the voicemail components reach 100% accuracy without ever misclassifying a live human, where Asterisk's AMD misclassifies 30 to 66% of them. For the AI callee, acoustic features do not transfer across recording conditions, but speech timing does, separating AI from human at 0.952 ROC-AUC, because synthesised speech leaves far shorter gaps between words. On production traffic, separating a live person from any automated far end reaches F1 0.979, and only 2.9% of live people are given an automated label, the error that matters most for an agent acting on the result.