We’ve seen some success utilizing AI when conducting textual speech pattern analysis for real-world open-source intelligence (OSINT) investigations. For example, we have text data of a target in our possession, such as social media posts or emails, and we would like to determine if the same writing style matches the style of anonymous actors. We fed the AI with the source text data as a sample and provided several social media posts from various entities. The AI responded if a text pattern was found based on the original source text.
To test the accuracy of this method, we decided to test using our own social media posts, below is the prompt we started with:
Please analyze and match the speech pattern from the example I provide below. I will then give you several options that best match the speech pattern provided in the example. [example]
We then gave the AI three social media posts to choose from (with different authors, of course), with one option being our own post from a year prior.

In the case mentioned above, the AI system (OpenAI) was able to accurately identify the original text among a group of options, demonstrating its ability to analyze and compare speech patterns. This kind of analysis can be beneficial in OSINT investigations, where the goal is to gather and analyze information from various sources.
For example, text analytics and pattern analysis can be used to verify the authenticity of written communications, such as emails or social media posts, by comparing the language and style used to known examples of an individual’s writing. It can also be used to identify the authors of anonymous documents or to track the spread of misinformation online.