RESEARCH
Signals, systems, and field notes for AI influence defense.
Published threat investigations, peer-reviewed detection methods, and operationally validated research.
Research, reporting, COVERAGE, and institutional context
Media Coverage
External reporting
NEWSGUARD
dotNex investigation featured in NewsGuard’s Foreign Influence Risk Briefing
dotNex’s investigation into AI-generated military personas on TikTok was featured in NewsGuard’s Foreign Influence Risk Briefing — a publication tracking active foreign influence threats for news and information consumers. The briefing highlighted dotNex’s findings on a coordinated network of fake U.S. soldier accounts operating during the Iran conflict.
STRAIGHT ARROW NEWS
Coverage of dotNex’s TikTok influence operation findings
dotNex uncovered a network of ~110 AI-generated military personas on TikTok accumulating 3.9M views — coordinated accounts posing as U.S. soldiers, with Chinese-language metadata embedded in the content. Straight Arrow News reported on the investigation and its findings.
CNN
CNN reference to dotNex’s CEO influence defense research
CNN cited dotNex CEO Luca Luceri’s work on AI-driven influence operations, including how foreign actors deploy synthetic personas and coordinated networks to manipulate public opinion on social media platforms.
FINANCIAL TIMES
Coverage of dotNex researchers’ investigation into the 2025 Canadian Election
dotNex researchers identified coordinated inauthentic activity targeting the 2025 Canadian federal election — networks of accounts amplifying partisan narratives across social media. The Financial Times covered the investigation and its implications for election integrity.
TIME
dotNex's CEO cited in TIME on Meta's fact-checking rollback
TIME magazine cited dotNex CEO Luca Luceri in its coverage of Meta's decision to replace professional fact-checkers with a Community Notes model on Facebook and Instagram — raising concerns that the change could accelerate the spread of misinformation at scale.
EUROPEAN COMMISSION — JOINT RESEARCH CENTRE
dotNex co-founder’s research cited in EC policy report on information warfare
dotNex co-founder Federico Cinus’s work on cross-platform coordinated inauthentic activity was cited in “Fractured Reality” — a policy report by the Joint Research Centre, the European Commission’s science and knowledge service, on how democracies can counter global information operations.
Publications
Related research
ICWSM 2026 · CONFERENCE PAPER
Coordinated Inauthentic Behavior on TikTok
The first systematic study of coordinated inauthentic behavior in a video-first ecosystem. Introduces new detection challenges specific to TikTok and proposes a multimodal detection framework.
THE WEB CONFERENCE 2026 · CONFERENCE PAPER
Emergent Coordinated Behaviors in Networked LLM Agents
Shows that networks of LLM agents spontaneously develop coordinated influence strategies with minimal human direction — providing empirical grounding for AI-based threat models.
AAAI 2025 · CONFERENCE PAPER
IOHunter: Graph Foundation Model to Uncover Online Information Operations
Introduces IOHunter, a graph-based foundation model for detecting influence operations at scale across platforms, fusing language models and network structures with a cross-attention mechanism.
THE WEB CONFERENCE 2025 · CONFERENCE PAPER
Exposing Cross-Platform Coordinated Inauthentic Activity in the Run-Up to the 2024 US Election
Maps how coordinated inauthentic actors coordinate and synchronize across platforms ahead of high-stakes events — fusing behavioral signals with information diffusion across multiple channels and media.
FIRST MONDAY 2024 · JOURNAL
Uncovering Coordinated Cross-Platform Information Operations Threatening the 2024 US Presidential Election
Documents a real-world coordinated influence campaign spanning multiple platforms during the 2024 election cycle — combining network analysis with behavioral fingerprinting.
THE WEB CONFERENCE 2024 · CONFERENCE PAPER
Unmasking the Web of Deceit: Uncovering Coordinated Activity to Expose Information Operations on Twitter
Introduces a network-based coordination detection method that surfaces information operations without relying on ground truth data — enabling proactive rather than reactive detection.
EPJ DATA SCIENCE 2023 · JOURNAL
Exposing Influence Campaigns in the Age of LLMs: A Behavioral-Based AI Approach
Detects state-sponsored troll accounts through behavioral cues alone — no text or language signals. Because it ignores content entirely, the approach remains effective even as LLMs make linguistic mimicry trivial .
ICWSM 2020 · CONFERENCE PAPER
Detecting Troll Behavior via Inverse Reinforcement Learning
Use inverse reinforcement learning to recover the reward structure driving the behavior of state-sponsored human operators — enabling detection of intent, not just action.