The social networking website has announced its move to work with the startup - which focuses on machine-learning technology - to use its Geometric Deep Learning tech "to detect network manipulation".
In a blog post Twitter's CTO Parag Agrawal wrote: "The result is the ability to analyze very large and complex datasets describing relations and interactions, and to extract signals in ways that traditional ML techniques are not capable of doing."
It was previously reported by TechCrunch that the Fabula AI system can detect fake news 93% of the time.
Agrawal added that Twitter will be in a better position to "improve the health of the conversation" and "help people feel safe on Twitter" thanks to the new move.
Meanwhile, the company previously tried out changing how users can engage with tweets.
Twitter's Sara Haider said: "Our goal is to make conversations easier to understand, and with the new threaded layout, people told us they found conversations easier to follow and liked seeing more of the conversation.
"Hiding engagements in replies resulted in the most mixed reviews: While some found themselves focusing more on the conversation as opposed to the metrics, others felt that the extra step to tap made it harder to Like and Retweet replies.
"Some felt that replies can get crowded, especially in long conversations. Lengthy threads become harder to read due to indentation and too many "Show more" buttons.
"Most people prefer the twttr reply highlighting, but also expressed visibility issues in dark mode. There was overall support for replying on twttr, specifically with the ability to reply in one place within a conversation; yet at times, it was confusing who you're replying to."