How Artificial Intelligence Is Being Used to Hide Misconduct, Fabricate Reality, and Bury the Truth (Part 3)

Decades ago, digital investigations largely focused on discovering what happened through traditional evidence sources: emails, text messages, documents, system logs, browser history, and metadata. 

The challenge was finding relevant evidence, preserving it defensibly, and reconstructing events accurately. 

That challenge is changing. 

Artificial intelligence is rapidly transforming litigation, digital forensics, internal investigations, and E-Discovery. Legal teams are using AI to review documents faster. Investigators are leveraging it to identify patterns and anomalies. E-Discovery professionals and digital forensic analysts are using AI-assisted workflows to reduce review burden and surface meaningful evidence more efficiently. 

At its best, AI helps uncover truth. But AI is also changing something far more concerning. 

It is changing how misconduct itself occurs. 

The same technology helping businesses, legal teams, and investigators move faster is also helping fraudsters, insiders, manipulators, and bad actors operate with greater sophistication, speed, and scale. The most powerful investigative technology ever created may also become the most powerful deception tool ever deployed. 

This article is the third installment in our AI on Trial series. In Part One, we explored how AI is transforming digital investigations by accelerating insight and improving efficiency. In Part Two, we examined how AI-generated evidence is reshaping questions of authenticity and admissibility. 

Part Three explores the next evolution of this challenge: What happens when AI is no longer just analyzing evidence or creating evidence, but actively helping conceal misconduct? 

The answer presents one of the most significant challenges modern litigation teams may face. 

AI-Assisted Misconduct Is Already Here 

The concern is no longer theoretical. 

AI is already being used to create: 

  • False narratives 
  • Fake communications 
  • Altered images 
  • Synthetic voice messages 
  • Fabricated documents 
  • Misleading exhibits 
  • Deepfake video 
  • Synthetic screenshots 

These risks extend across both civil and criminal matters. Potential implications include: 

  • Fraud 
  • Theft 
  • Harassment 
  • Trade secret disputes 
  • Internal misconduct 
  • Corporate investigations 

What once required advanced technical skill can now be done with consumer-grade tools in minutes. 

Fraud has entered the generative era. 

That changes the threat landscape dramatically. The barrier to deception has never been lower. 

Case Example: In re Morgan & Morgan / MX2.law AI Citation Sanctions Matter
U.S. District Court for the District of Wyoming (2025)
Judge Kelly Rankin 

Early legal examples already demonstrate how AI can introduce serious risk into litigation workflows. 

In 2025, Judge Kelly Rankin of the U.S. District Court for the District of Wyoming sanctioned attorney T. Michael Morgan after motions in limine were filed containing nine case citations, eight of which did not exist. 

The false citations were hallucinated by the firm’s internal AI platform, MX2.law. Morgan admitted he did not review the filings before they were submitted under his signature. The court found a violation of Rule 11 and imposed sanctions. 

This matter highlights one form of AI risk: overreliance without verification. 

But a more troubling threat is emerging.  

Not accidental AI error. Intentional AI-assisted duplicity. 

Case Example: When AI Conversations Become Evidence — Fortis Advisors LLC v. Krafton, Inc. 

One of the most significant AI-related discovery decisions to date is Fortis Advisors LLC v. Krafton, Inc., decided by the Delaware Court of Chancery in 2026. The case arose from a $250 million earnout dispute following Krafton’s acquisition of video game developer Unknown Worlds. During litigation, the court considered evidence showing that Krafton’s CEO had used ChatGPT to develop a strategy for avoiding the earnout payment. The AI prompts and responses included recommendations for controlling public messaging, securing leverage over the acquired company, and preparing for potential litigation. According to the court, many of those recommendations were later reflected in Krafton’s actions, making the AI conversation highly relevant to questions of intent, motive, and corporate decision-making. 

The significance of Fortis extends far beyond the underlying business dispute. For perhaps the first time, AI prompts and chatbot outputs became central evidence of a party’s state of mind rather than simply a productivity tool. The decision signals that interactions with generative AI may be discoverable when they influence business decisions or litigation strategy. For litigators and forensic professionals, Fortis marks an important turning point: AI-generated content is no longer just something we analyze—it may itself become evidence. The prompt, not just the document, may ultimately tell the most important part of the story.  

AI-assisted deception can: 

  • Distort timelines 
  • Manufacture intent 
  • Create false context 
  • Influence perception 
  • Coach bad behavior or manipulate harmful behavior

And in litigation, perception often matters almost as much as fact. 

The Rise of Narrative Engineering 

The greatest danger may not be fake evidence alone. The greater threat is narrative engineering. 

Narrative engineering is the deliberate use of AI to construct a believable, but false, version of reality using synthetic artifacts. 

AI can now: 

  • Fabricate conversations 
  • Simulate emotional tone 
  • Generate fake corroboration 
  • Create documents that never existed 

This changes the nature of deception. 

Historically, fraud often involved a single manipulated document or isolated false statement. Today, AI allows bad actors to fabricate entire ecosystems of credibility. 

A fake text thread. A fake email chain. A fake voicemail. Supporting screenshots. Corroborating explanations. 

Instead of forging one exhibit, AI makes it possible to manufacture an entire narrative designed to influence decision-makers. 

That is extraordinarily powerful, and extraordinarily dangerous. 

“Artificial intelligence can be a powerful tool, but lawyers abandon their responsibilities when they rely on AI-generated content without verification.” – Chief Judge P. Kevin Castel, Mata v. Avianca, Inc. 

The Hidden Risk: AI Helps Conceal Intent 

Traditional digital investigations typically look for clear indicators of activity. 

Investigators analyze: 

  • Transfers 
  • Access logs 
  • Deletions 
  • File artifacts 
  • Communication records 
  • Metadata 

These indicators remain critical. But AI changes where intent may live. 

Planning may occur entirely inside prompts. No native document may ever exist. Intent may be synthesized instantly and disappear just as quickly. 

This creates a profound shift in investigative strategy. The evidence may no longer live in the file. It may live in: 

  • Prompt histories 
  • AI outputs 
  • Browser artifacts 
  • Contextual metadata 
  • Cloud synchronization records 

That changes the investigative playbook significantly. Instead of simply asking what happened, investigators increasingly need to understand how information came into existence in the first place. 

The New Investigative Questions 

This new environment creates entirely new forensic questions. Investigators increasingly need to determine: 

  • Was AI used? 
  • What percentage of content was human-created versus AI-generated? 
  • Which platform or model was used? 
  • What content was synthetic? 
  • Who generated it? 
  • Was it altered after creation? 
  • What artifacts remain? 

These are no longer fringe-case questions. They are rapidly becoming central to digital investigations. 

Case Examples: Mata v. Avianca, Inc. 22-cv-1461 (PKC) United States District Court, Southern District of New York (2023) Judge P. Kevin Castel 

The case began as a personal injury action filed by plaintiff Roberto Mata against Avianca, arising from injuries allegedly sustained during an international flight. During briefing on Avianca’s motion to dismiss, plaintiff’s counsel submitted an opposition memorandum citing multiple judicial opinions that purportedly supported their legal arguments. 

The problem: several of those cited cases did not exist. 

When opposing counsel and the court attempted to locate the authorities, they discovered that six cited cases were entirely fictitious. The citations, quotations, and legal analyses had been generated by ChatGPT and were submitted without independent verification. 

Plaintiff’s counsel later admitted he had used ChatGPT to assist with legal research and relied on the platform’s false assurances that the cases were authentic. The AI system even generated fabricated case summaries and false quotations when prompted to verify the citations. Judge P. Kevin Castel issued a sanctions order in June 2023, finding that counsel acted in bad faith by submitting non-existent authorities and failing to conduct a reasonable inquiry into the law before filing. The court imposed a $5,000 sanction and required the attorneys to notify each judge falsely identified as authoring the fabricated opinions. 

The decision became a landmark warning to the legal profession about the risks of unverified AI-generated content in litigation. While the case involved negligent reliance on generative AI rather than intentional fraud, it underscored a broader concern now facing litigators, judges, and forensic investigators: AI-generated content can appear highly credible while being entirely false. 

The case established an early but important principle for AI use in legal practice: attorneys may use AI tools, but they remain fully responsible for verifying the accuracy, authenticity, and defensibility of anything submitted to a court. 

The questions are expanding. 

Where was this generated? What metadata supports authenticity? What technical artifacts validate provenance? What evidence contradicts the narrative being presented? 

These are becoming core questions in modern litigation. 

The Future of Digital Evidence 

Digital evidence is entering a new era. The future will increasingly involve: 

  • AI-generated communications 
  • Multimodal synthetic evidence 
  • AI-assisted decision making 
  • Hybrid human/AI authorship 

This introduces a new challenge for courts, counsel, and forensic investigators. 

As synthetic media improves, authentic evidence becomes easier to challenge. This phenomenon is often called the Liars’ Dividend. A “real” voicemail can be dismissed as fake. An authentic screenshot can be challenged as AI-generated. A legitimate recording can suddenly require extensive forensic validation. 

This changes everything. 

Investigators will increasingly need to: 

  • Validate authenticity 
  • Trace provenance 
  • Analyze prompt histories 
  • Separate reality from synthetic narrative 

“AI-generated content can be weaponized to impersonate individuals, commit fraud, and undermine trust in digital evidence.” — Federal Bureau of Investigation 

Looking Ahead 

Artificial intelligence can help uncover truth. But it can also: 

  • Obscure it 
  • Automate deception 
  • Scale misconduct 

That is the new reality facing litigation teams, digital forensic experts, and investigators. 

The future challenge of digital forensics may not be finding evidence. It may be determining whether the evidence, or the story surrounding it, was ever real to begin with. In an era of synthetic narratives, truth will increasingly depend on defensible collection, forensic validation, and trusted expertise. 

Veracity Forensics helps legal teams navigate emerging evidentiary risks with transparency, technical rigor, and defensibility at the forefront.