UK Startup Fabula AI Spots Fake News by Analyzing Social Spread Patterns
Fabula AI bypasses the need to understand text content by using geometric deep learning to study how information spreads across social networks. The startup claims fake and real news exhibit distinct propagation patterns that its algorithms can easily detect.
UK startup Fabula AI develops a new artificial intelligence method to tackle the ongoing disinformation crisis on social media platforms. Instead of attempting to read and understand text like traditional content moderation tools, this approach relies on a patented form of geometric deep learning to analyze complex network data.
The technology focuses entirely on how information propagates across social networks rather than evaluating the content itself. Co-founder Michael Bronstein explains that fake news and genuine news exhibit distinctly different spreading patterns, allowing the algorithm to identify falsehoods by simply mapping the behavior of the users sharing them.
Fabula AI targets both intentional disinformation and unintentional misinformation under its detection umbrella. By operating in the non-Euclidean space of social graphs where traditional machine learning struggles, this breakthrough method offers platforms a faster, context-free way to flag fabricated content before it causes widespread harm.