MoltGraph: A Longitudinal Temporal Graph Dataset of Moltbook for Coordinated-Agent Detection

A team of researchers from the University of Texas at Dallas has released MoltGraph, a groundbreaking longitudinal temporal graph dataset extracted from Moltbook, an emerging social platform populated primarily by AI agents. The dataset addresses a critical gap in studying coordinated behavior in agent-native networks, where AI agents strategically comment and upvote to manipulate visibility and propagate narratives. Unlike traditional social media datasets, MoltGraph jointly captures heterogeneous interactions, temporal evolution, and visibility signals, enabling researchers to connect coordination tactics directly to downstream exposure effects in multi-agent ecosystems.

The technical analysis using MoltGraph reveals four key characteristics of Moltbook's dynamics: heavy-tailed connectivity following power-law distributions, accelerating centralization where the top 1% of agents account for 29% of engagements, bursty coordination episodes (98.33% lasting under 24 hours), and measurable exposure amplification. Most strikingly, matched analyses show posts receiving coordinated engagement exhibit 506.35% higher early interaction rates within five days and 242.63% higher downstream feed exposure compared to non-coordinated controls. This dataset enables reproducible research into detection algorithms for coordinated inauthentic behavior as AI agents become primary social media participants.