Cite Tian, L., Booth, E., Bailo, F., Droogan, J., & Rizoiu, M.-A. (2025, February 7). Before It’s Too Late: A State Space Model for the Early Prediction of Misinformation and Disinformation Engagement. WWW ’25: Proceedings of the ACM Web Conference 2025. 2025 ACM Web Conference, Sydney. https://doi.org/10.1145/3696410.3714527

In today’s fast-moving digital world, false information can spread like wildfire, influencing public opinion and even destabilising democracies. A new study introduces IC-Mamba, an AI-powered model designed to predict how misinformation and disinformation gain traction online—before it’s too late.

  • Early Detection of Viral Misinformation
    IC-Mamba can forecast how misinformation spreads within the first 15-30 minutes of a post going live, allowing for rapid intervention.

  • Advanced AI for Social Media Analysis
    Unlike traditional models, IC-Mamba processes irregularly sampled social media data to better understand engagement trends over time.

  • Improved Accuracy Over Existing Models
    By integrating interval-censored data and state space modeling, IC-Mamba improves prediction accuracy by 4.72% over state-of-the-art models across multiple platforms.

  • A Tool for Proactive Action
    This technology enables researchers, platforms, and policymakers to identify harmful narratives early, helping to prevent large-scale disinformation campaigns.

As misinformation becomes more sophisticated, AI-driven solutions like IC-Mamba could play a crucial role in safeguarding online spaces and ensuring the integrity of public debate.