arXiv:2606.03686v1 Announce Type: new Abstract: We present DeepSpeak-Agentic, a dataset of videos comprising over 37 hours of semi-structured conversations between a human and an embodied AI agent. We use this dataset to evaluate the automatic forensic identification (audio, video, or text) of AI agents, study the nature of human-agent interactions, and provide a benchmark for future advances in the large-language models and AI-generated voices and faces that power embodied AI agents. We also contribute a scalable data-capture system that creates agents, automatically pairs them with human cro
The release of a comprehensive dataset for human-AI agent interaction reflects the rapid development and increasing complexity of AI agentic systems and the need for standardized evaluation.
This dataset provides a critical benchmark for evaluating AI agent performance, especially in forensic identification and understanding human-agent dynamics, which is crucial for safety and efficacy.
The ability to forensically identify AI agents across modalities will enhance trust and accountability while accelerating the development of more sophisticated and human-like embodied AI.
- · AI researchers
- · Embodied AI developers
- · AI ethics and safety organizations
- · Malicious actors deploying undetectable AI agents
Improved detection capabilities for deepfakes and AI-generated content across audio, video, and text.
Accelerated development of more sophisticated and adaptable embodied AI agents due to better evaluation tools.
Enhanced regulatory frameworks and public trust in human-AI interactions as transparency and identification become more robust.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI