Real time detection of coordinated bots on Bluesky

Summary

This paper addresses the detection of coordinated bot activity on Bluesky, a decentralized social media platform that has received comparatively little attention in the bot-detection literature. The authors argue that misinformation gains credibility through viralization, and that viralization is often manufactured by coordinated bot networks rather than emerging organically. They propose a real-time detection approach aimed at identifying such coordinated behavior as it unfolds, distinguishing malicious coordination from benign organic activity.

Key Contributions

• Extends coordinated inauthentic behavior detection to Bluesky, an emerging decentralized platform underrepresented relative to X, Instagram, and TikTok.
• Frames detection around coordination in real time, rather than retrospective post-hoc analysis or per-account bot scoring.
• Articulates a conceptual distinction between organic coordination and malicious coordinated bot activity as the basis for detection.

Methods

• Case study on the Bluesky platform.
• A real-time detection pipeline for coordinated bot networks (specific algorithmic details are not available from the abstract).

Findings

• Specific empirical results are not reported in the available abstract.

Connections

This work sits alongside other recent efforts targeting Bluesky as a new venue for coordinated behavior research, notably Graham2025-gp and Hurcombe2025-cs. Its real-time framing connects to streaming and online detection approaches such as Iannucci2025-eg and Minici2024-tf, while the broader conceptual move of separating organic from malicious coordination echoes Luceri2025-tr and Graham2026-fb.

Podcast

A research-radio episode discusses this paper: Listen