Wolf of Wall Street or Shepherd of Sarajevo? - Streamlining Document Verification with AI
Validating the authenticity of documents is a critical, albeit challenging, task. With the power of AI, we're enhancing the process of linguistic analysis to verify the origins and authors of key correspondences.
by Kyrill Lapin
Navigating the Document Deluge
When diving into the deep sea of corporate due diligence, we frequently surface with various documents - contracts, correspondences, letters - from potential partners or alleged investors. Verifying their authenticity can be a difficult task. This becomes especially challenging when we need to determine if a document is indeed authored by a specific industry professional or a native English speaker.
Take an example where we receive a batch of letters supposedly authored by a high-profile Wall Street investor. To an experienced analyst, something might seem off. Maybe the letters are too sterile, lacking the splash of Wall Street lingo that peppers the communication of a seasoned investor. Or perhaps, American idioms are oddly absent, replaced by structures more common to British English. Manually pinpointing these nuances can be daunting — akin to finding a needle in a haystack.
AI dives into the intricacies of language use, noting stylistic patterns, vocabulary, and grammar structures typical of native speakers or specific industry professionals. By juxtaposing these factors with known linguistic profiles, AI gives us a reasonable estimate of a document's authenticity, suggesting whether these letters were truly composed by the supposed author or someone in disguise.
Enhancing Linguistic Analysis
This capability is a game-changer for risk mitigation. It helps us ascertain potential partners' credibility without stirring potential discord by direct contact. In fraud recovery cases, it aids in determining the authenticity of correspondences and collecting evidence. Plus, it's not just about precision - AI ensures cost and time efficiency as well. It eliminates the need for expensive professional linguistic analysis and saves a significant amount of time by quickly processing large volumes of documents, empowering us to focus on strategic decision-making.
Some examples from our recent work include:
A supposed bank reference letter was analyzed, with AI confirming it could not be authentic based on linguistic and stylistic discrepancies.
In another case, an email from a purported supplier was discovered to be a fabrication by a client's manager, who had established a fake identity.
An alleged email from a UK-based investor was found to be written by someone from Eastern Europe, debunked through a linguistic analysis that revealed inconsistencies with UK English.
Countering AI-Generated Forgeries
In an era where AI is becoming ubiquitous, it's crucial to note that those with malicious intent could potentially use AI to craft convincing forgeries. However, the evolution of AI is a double-edged sword. These AI-generated correspondences tend to be too clean, lacking the human touch of casual language use or industry-specific jargon. AI's tendency to stick too closely to grammatical rules can also be a giveaway, as humans often employ creative license in their writing.
Again, a well prompted AI can help to detect such patterns and detect attempts to mask the true identity of the author. So, even as the forgeries become more realistic, our methods of detection are constantly improving, ensuring that we can still discern truth from fiction.
