AI & PI: What Does the Future Hold? Why Licensed Investigators Become More Valuable When Anything Can Be Faked

By Alexander Maynfeld, Director of Investigations, MAYN Investigations (Florida agency license A3400126). A perspective for Florida litigators, family lawyers and paralegals.

In September 2025, a California judge threw out an entire lawsuit because the plaintiffs had filed deepfake videos of a witness as evidence. The judge noticed that the “witness” barely changed expression, her mouth did not match her words, and the clip seemed to loop. When the court compared the clips with an authentic recording of the same woman, it found they were not the same person. ^[7][8]

That case, Mendones v. Cushman & Wakefield, is a preview. Fake evidence is no longer a hypothetical for a law review article. It is showing up in filings, and for now it is being caught mostly because someone looked closely and compared it against something real.

I run a licensed private investigation agency in Florida. My view of where this goes is simple, and it is the reason I wrote this piece:

When anything can be manipulated, the people who can be held accountable for the truth become the most valuable people in the case.

A license does not give an investigator magic powers to spot a deepfake. What it gives is something a file can never have: a regulated, named, accountable human being who gathered the evidence at the source and can be put under oath and cross-examined about it. I call that the Accountability Premium, and I believe it will define the next decade of litigation.

The numbers: fakery is getting cheap, fast

The cost of producing convincing fake media has collapsed, and fraud figures are following.

• The FBI’s 2025 Internet Crime Report recorded about $20.9 billion in reported cybercrime losses, up 26% from 2024. For the first time in its nearly 25-year history, the report included a section on artificial intelligence: 22,364 complaints and nearly $893 million in reported losses. ^[2][3]
• Deloitte’s Center for Financial Services projects that generative AI could enable U.S. fraud losses to reach $40 billion by 2027, up from $12.3 billion in 2023, a 32% compound annual growth rate. ^[1]
• FinCEN issued a formal alert in November 2024 (FIN-2024-Alert004) after seeing an increase in suspicious activity reports describing deepfake media, often used to create fraudulent identity documents and get past identity checks. ^[4]
• Entrust’s 2025 Identity Fraud Report, based on its identity verification data, found a deepfake attempt every five minutes in 2024, and digital document forgeries up 244% year over year. ^[5]

And the headline case every corporate client should know: in early 2024, an employee of the engineering firm Arup in Hong Kong joined a video call with what appeared to be the company’s CFO and other colleagues. They were AI-generated. He made 15 transfers totaling HK$200 million, about US$25.6 million. ^[6]

The point for litigators is not the size of the losses. It is what these numbers say about evidence. If a fake CFO can pass a live video call, a fake text thread, a fake voicemail or an edited dashcam clip can pass a quick look in discovery.

Deepfakes have already reached the courtroom

Two cases show the shape of the problem.

Family law. Byron James, a Dubai-based family lawyer, described a UK custody dispute in which an audio recording was offered to show his client, a father, making threats. James reported that the recording had been doctored, and that experts were able to show the manipulation by examining the original file and its metadata. ^[9] Custody and domestic violence matters are where this risk is highest, because one recording can decide who sees a child.

Civil litigation. In Mendones, the Alameda County Superior Court found that videos and images submitted with a summary judgment motion were deepfakes or altered, reviewed the metadata, and issued terminating sanctions on September 9, 2025. ^[7][8]

There is a second danger, and it may be bigger. Law professors Robert Chesney and Danielle Citron called it the “liar’s dividend”: once the public knows deepfakes exist, people caught on real recordings can claim the real thing is fake. ^[10] Every authentic video your client depends on is now open to the accusation “that’s AI.” Answering that accusation takes a witness who can explain where the recording came from, who handled it, and why it is what it claims to be.

The rules of evidence are still catching up

Federal rulemakers are taking this seriously, but they have not finished.

The federal Advisory Committee on Evidence Rules has spent several years studying AI. It identified two separate problems: machine-generated evidence that would need expert-level reliability if a human offered it, and audiovisual evidence that is not authentic because it is a hard-to-detect deepfake. ^[11]

For the first problem, the Committee proposed a new Federal Rule of Evidence 707, which would apply the Rule 702 reliability standard to machine-generated evidence offered without an expert. It was published for public comment from August 2025 to February 2026. In May 2026, the Committee decided not to recommend final approval. It revised the proposal and set it for further study alongside the deepfake authentication problem, including a possible new Rule 901(c), at its fall 2026 meeting. ^[11][12][13]

So as of this writing, there is no special federal deepfake rule. Authentication still runs through Rule 901, and in Florida through section 90.901 of the Evidence Code, which requires “evidence sufficient to support a finding that the matter in question is what its proponent claims.” ^[15]

That standard was written for a world where faking a recording was hard. It is a low bar. In practice, that means the quality of the foundation witness matters more than ever. Who collected it? From where? How was it preserved? Can that person explain it under oath?

Lawyers have already learned the AI lesson the hard way

The legal profession got its own warning in 2023. In Mata v. Avianca, a federal judge in New York sanctioned lawyers who filed a brief citing cases that did not exist. The citations had come from ChatGPT, and the court imposed a $5,000 penalty. ^[14]

The Florida Bar responded quickly. Florida Bar Ethics Opinion 24-1 (January 2024) allows lawyers to use generative AI, but says they remain responsible for their work: protect client confidentiality, supervise the tool, and verify what it produces. ^[16]

That same principle, verify, don’t trust, is exactly what evidence now requires. And verification is what licensed investigators do.

Florida is already legislating against synthetic media

Florida has moved on several fronts:

• Section 836.13, Florida Statutes, makes it a crime to create or share certain “altered sexual depictions” of identifiable people. ^[17]
• Brooke’s Law (CS/CS/HB 1161, chapter 2025-133), signed in June 2025, requires covered online platforms to set up a process for removing nonconsensual altered sexual depictions, and to remove them within 48 hours of a valid request. ^[18][19]
• Section 106.145, added in 2024, requires disclaimers on political ads that use generative AI in certain ways. ^[20]

The trend is clear. Lawmakers are treating synthetic media as a real harm. Courts will follow with sharper questions about where evidence came from.

Why the license matters more, not less

In Florida, private investigation is a regulated profession under Chapter 493. An individual Class “C” investigator must have two years of qualifying experience or training and pass a state exam, and must own or work for a licensed Class “A” agency. ^[21] Licensees answer to the Florida Department of Agriculture and Consumer Services, which can discipline them for things like betraying client confidences, using tracking devices illegally, or assisting violations of the law. ^[22] Anyone can check a license for free on the FDACS website. ^[23]

Florida also sets hard limits on how evidence is gathered. Recording a private conversation generally requires every party’s consent under section 934.03, and section 934.06 bars unlawfully intercepted wire and oral communications from being used as evidence. Placing a tracking device on someone’s vehicle without consent is a crime under section 934.425. ^[24][25] A licensed investigator works inside those lines because their license depends on it.

Here is what that means when evidence is attacked as fake:

1. Original-source collection. I pull records from the source: the clerk, the property appraiser, the state’s corporate filings, the business that holds the video. I don’t rely on a screenshot someone forwarded. Evidence collected at the source is far harder to fake than evidence that arrived in a chat.
2. Firsthand observation. Surveillance done by a licensed investigator is documented by a person who was there, with time, place and conditions noted. That person can be deposed.
3. Chain of custody. Original files are preserved with their metadata intact and logged from collection to production. When the other side says “that’s AI,” there is a documented history to answer with.
4. A witness who can be cross-examined. A video cannot explain itself. An investigator can. The federal Committee made the same point in its Rule 707 work: “A human expert can be cross examined,” while with machine output, “the inability to cross-examine is a concern.” ^[11]
5. Accountability. A licensed investigator has a regulator, a license number and a career on the line. An anonymous source, an unlicensed “consultant,” or an AI tool has none of those.

A file cannot be cross-examined. An investigator can.

How a responsible PI uses AI without letting it decide

I am not anti-AI. Used well, it makes an investigation faster. The rule I follow is simple: AI can point. A human must prove.

• Triage, not truth. AI tools can help sort large volumes of open-source information, organize social media content, transcribe recordings and flag leads. Every lead is then checked against original sources before it appears in a report.
• No AI-made evidence. I don’t use AI to create, enhance or “clean up” images, audio or video that will be offered as evidence. Enhancement invites the exact challenge you are trying to avoid.
• Confidentiality. Client information doesn’t go into public AI tools. That matches the approach Florida Bar Opinion 24-1 asks of lawyers. ^[16]
• Disclosure. If a tool was used in a way that matters to the work, the report says so.

What attorneys should do now

1. Ask where every key exhibit came from. If the answer is “the client sent it,” get the original file, not a screenshot or forwarded copy.
2. Preserve originals early. Send preservation letters for video, phone data and accounts before routine deletion happens.
3. Lock down metadata. Collect native files with their metadata. Keep a written chain of custody from the start.
4. Build the foundation witness now. Identify who can testify to how each recording or record was created and collected. That person should be someone who can survive cross-examination.
5. Scrutinize the other side’s media. Ask for native files and metadata. Compare questioned recordings to known authentic material. Bring in a qualified forensic examiner when the stakes justify it.
6. Verify your investigator’s license. Run it on the FDACS lookup before the work starts, and confirm they know Florida’s recording and tracking laws. ^[23]
7. Plan for “that’s AI.” Assume authentic evidence will be attacked as fake, and build the record so you can prove it is real.
8. Set an AI policy for your team. Follow Opinion 24-1: protect confidential information and verify everything a tool produces. ^[16]

What the future holds: my predictions for 2026 to 2030

These are my opinions, based on where the rules, the fraud data and the cases are heading. They are not guarantees.

1. “Is this real?” becomes a standard question. I expect authenticity challenges to digital media to become routine in family, employment and business litigation, the way spoliation questions did after e-discovery arrived.
2. Federal rules will change, and states will follow. I expect the federal Committee to settle on some version of a reliability rule for machine output and a deepfake authentication rule within the next few years, with Florida considering its own approach after that.
3. Provenance will beat detection. Detection tools will keep improving, and so will the fakes. The winning approach will be proving where evidence came from: original sources, preserved metadata and documented custody.
4. The human witness becomes a premium. As machine-made content gets cheaper, evidence that comes with an accountable person attached becomes more valuable. That is the Accountability Premium.
5. Investigator standards will rise. Courts and clients will expect investigators to document collection and custody of digital evidence with the same care that forensic labs do today.

The bottom line

AI won’t replace investigators. It will raise the stakes for them. When anything can be faked, what a case needs most is someone who was there, who gathered the evidence lawfully at the source, who kept it intact, and who can stand behind it under oath. That’s what a license represents: not a promise of perfection, but a person who can be held accountable.

Evidence wins cases. In the age of AI, verified evidence wins cases.

This article is general information for attorneys and legal professionals. It is not legal advice. Predictions are the author’s opinion. Decisions about admissibility, discovery and strategy belong to counsel.

References

1. Deloitte Center for Financial Services, “Generative AI is expected to magnify the risk of deepfakes and other fraud in banking,” Deloitte Insights. https://www.deloitte.com/us/en/insights/industry/financial-services/deepfake-banking-fraud-risk-on-the-rise.html
2. FBI National Press Office, “Cryptocurrency and AI Scams Bilk Americans of Billions,” April 6, 2026. https://www.fbi.gov/news/press-releases/cryptocurrency-and-ai-scams-bilk-americans-of-billions
3. FBI Internet Crime Complaint Center, 2025 IC3 Annual Report. https://www.ic3.gov/AnnualReport/Reports/2025_IC3Report.pdf
4. FinCEN, “FinCEN Issues Alert on Fraud Schemes Involving Deepfake Media Targeting Financial Institutions” (FIN-2024-Alert004), November 13, 2024. https://www.fincen.gov/news/news-releases/fincen-issues-alert-fraud-schemes-involving-deepfake-media-targeting-financial
5. Entrust, “Deepfake Attempts Occur Every Five Minutes Amid 244% Surge in Digital Document Forgeries,” November 19, 2024. https://www.entrust.com/company/newsroom/deepfake-attacks-strike-every-five-minutes-amid-244-surge-in-digital-document-forgeries
6. CNN, “Arup revealed as victim of $25 million deepfake scam involving Hong Kong employee,” May 16, 2024. https://www.cnn.com/2024/05/16/tech/arup-deepfake-scam-loss-hong-kong-intl-hnk
7. NBC News, “AI-generated evidence showing up in court alarms judges.” https://www.nbcnews.com/tech/tech-news/ai-generated-evidence-deepfake-use-law-judges-object-rcna235976
8. eDiscovery Today, “Deepfake Videos and Images Lead to Terminating Sanctions,” discussing Mendones v. Cushman & Wakefield, Inc., No. 23CV028772 (Cal. Super. Ct. Sept. 9, 2025). https://ediscoverytoday.com/2025/09/25/deepfake-videos-and-images-lead-to-terminating-sanctions-ediscovery-case-law/
9. The National, “‘Deepfake’ audio evidence used in UK court to discredit Dubai dad.” https://www.thenationalnews.com/uae/courts/deepfake-audio-evidence-used-in-uk-court-to-discredit-dubai-dad-1.975764
10. Robert Chesney and Danielle Keats Citron, “Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security,” California Law Review. https://www.californialawreview.org/print/deep-fakes-a-looming-challenge-for-privacy-democracy-and-national-security
11. Advisory Committee on Evidence Rules, Report (December 1, 2025). https://www.uscourts.gov/sites/default/files/document/2025-12-01_evidence_rules_committee_report.pdf
12. Advisory Committee on Evidence Rules, Report (May 2026). https://www.uscourts.gov/sites/default/files/document/advisory_committee_on_evidence_rules_may_2026.pdf
13. Advisory Committee on Evidence Rules, Agenda Book (October 2026). https://www.uscourts.gov/sites/default/files/document/2026-10-evidence-rules-agenda-book.pdf
14. Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023), Opinion and Order on Sanctions. https://www.nhd.uscourts.gov/sites/default/files/pdf/Mata-v-Avianca-sanctions-order.PDF
15. Section 90.901, Florida Statutes. https://www.flsenate.gov/Laws/Statutes/2026/90.901
16. The Florida Bar, Ethics Opinion 24-1 (January 19, 2024). https://www.floridabar.org/etopinions/opinion-24-1/
17. Section 836.13, Florida Statutes. https://www.flsenate.gov/Laws/Statutes/2026/836.13
18. Laws of Florida, chapter 2025-133 (CS/CS/HB 1161). https://laws.flrules.org/2025/133
19. Executive Office of the Governor, “Governor Ron DeSantis Signs Legislation to Protect Children Against Sexual Crimes,” June 2025. https://www.flgov.com/eog/news/press/2025/governor-ron-desantis-signs-legislation-protect-children-against-sexual-crimes
20. Section 106.145, Florida Statutes. https://www.flsenate.gov/Laws/Statutes/2026/106.145
21. Sections 493.6201 and 493.6203, Florida Statutes. https://www.flsenate.gov/Laws/Statutes/2026/493.6201 and https://www.flsenate.gov/Laws/Statutes/2026/493.6203
22. Section 493.6118, Florida Statutes. https://www.flsenate.gov/Laws/Statutes/2026/493.6118
23. FDACS Division of Licensing, Private Investigation Licenses. https://www.fdacs.gov/Business-Services/Private-Investigation-Licenses
24. Sections 934.03 and 934.06, Florida Statutes. https://www.flsenate.gov/Laws/Statutes/2026/934.03 and https://www.flsenate.gov/Laws/Statutes/2026/934.06
25. Section 934.425, Florida Statutes. https://www.flsenate.gov/Laws/Statutes/2026/934.425