A Perfectly Preserved Lie: Why Authentic Evidence Isn’t Always the Truth

A document can be authentic, admissible, and completely false. A photograph can be genuine, properly preserved, and misleading. And a fabricated document can have perfect forensic integrity.

How can all three things be true?

The answer lies in understanding something that attorneys and digital forensic experts sometimes approach from different angles: authenticity, admissibility, and trustworthiness are not the same thing. As artificial intelligence becomes increasingly involved in creating, modifying, and analyzing evidence, understanding those distinctions will become more important.

A perfectly preserved lie

Let’s start with a simple example.

Imagine I create a document today describing a meeting that never occurred, complete with a convincing title, attendee names, and detailed notes on decisions that were never made. Then I calculate its cryptographic hash value and preserve the document in a secure evidence repository.

Six months later, a forensic examiner retrieves the file, calculates its hash value, and confirms that it hasn’t changed. The examiner has established something important: the document has maintained its integrity since the original hash was calculated.

But the meeting still never happened.

Integrity is not the same thing as authenticity, and authenticity is not the same thing as truth. In digital forensics, a hash value can help establish that a file hasn’t changed. It cannot establish that the file was genuine when it was created.

Likewise, a document can be authentic as a record of what someone wrote without the statements within it being true.

What does authentication actually establish?

In federal litigation, authentication is addressed principally by Federal Rule of Evidence 901. The rule generally requires sufficient evidence to support a finding that an item is what its proponent claims it to be.

Notice what that does—and doesn’t—mean. Authentication doesn’t necessarily establish that a document’s contents are truthful, that a photograph provides complete context, or that a witness’s interpretation is correct. It establishes a sufficient foundation for identifying the evidence as what it is claimed to be.

If I can establish that a particular person sent an email on a specific date, I may have authenticated the communication. But what if the sender was lying? The email remains authentic as a communication. ts contents may still be false. And depending on why the email is being offered, its contents may raise separate questions under the hearsay rules.

Authentication is an important legal requirement. But it is not a guarantee of truth.

Admissibility is a different question

Even authenticated evidence isn’t automatically admissible. The Federal Rules of Evidence impose additional requirements. Evidence generally must be relevant. Under Rule 401, relevance concerns whether evidence has a tendency to make a consequential fact more or less probable.

Other rules may restrict admission because of hearsay, privilege, unfair prejudice, or other considerations. Expert testimony presents additional questions concerning qualifications, methodology, and reliability under Rule 702. These requirements aren’t interchangeable, and they don’t apply identically to every type of evidence. But collectively, they help determine whether evidence may be admitted.

Authentication is one requirement that may support admissibility. Relevance is another. Admissibility is the broader legal determination. And even when evidence is admitted, the factfinder may give it very little weight. That’s where the next distinction becomes important.

Admissibility doesn’t mean believability

Imagine a photograph showing two executives having lunch. It hasn’t been altered, the photographer can establish when and where it was taken, and the image is properly authenticated and admitted into evidence.

One party argues that it proves the executives were conspiring; the other argues they were simply having lunch. Both sides agree that the photograph is genuine, but they disagree on what it actually proves.

This is the difference between admissibility and evidentiary weight.

Admissibility determines whether the photograph may be considered. Evidentiary weight concerns how much significance the factfinder should assign to it. And trustworthiness concerns whether the evidence deserves reliance for the proposition being advanced.

The photograph might be completely trustworthy as evidence that two people met for lunch, but provide little support for an allegation of conspiracy. Evidence can be authentic, admissible, and trustworthy for one proposition while being unpersuasive for another. This is why context matters.

I’ve found it useful to think about evidence evaluation in four interconnected layers.

A framework for evaluating evidence:

1. Evidence characteristics: What is the evidence, where did it come from, and what does it purport to represent? This is where concepts such as integrity, fidelity, and provenance become important. Integrity concerns whether evidence has changed. Fidelity concerns whether it faithfully represents what it claims. Provenance concerns its origins and history. These are characteristics we can investigate.
2. Forensic assessment: What can we establish about those characteristics, and how certain are we? Confidence isn’t an inherent characteristic of a document. It describes the strength of our conclusions about the document. We might have high confidence that a file hasn’t changed, moderate confidence about its origin, and limited confidence about whether its contents accurately represent an event. A sound forensic assessment should identify both its findings and its limitations.
3. Legal admissibility: Does the evidence satisfy the applicable legal requirements for admission? This includes relevance, authentication where required, and other applicable evidentiary rules. The court’s role is to determine whether the legal requirements have been satisfied. That determination doesn’t necessarily resolve every factual dispute about the evidence.
4. Evidentiary weight: Assuming the evidence is admitted, how much should the factfinder rely upon it? Trustworthiness, credibility, context, and competing evidence become especially relevant at this stage. The same document may be persuasive for one purpose and unpersuasive for another.

These four layers are interconnected, but they aren’t identical. And confusing them can lead to mistaken assumptions about what forensic examinations actually establish.

Why AI makes these distinctions more important

Consider an AI-generated report summarizing a company’s financial transactions. The report may be unquestionably authentic as an output produced by a particular AI system. Its integrity may be established through reliable preservation procedures. But does it accurately represent the underlying transactions? Did the system omit important information? Misinterpret a transaction? Invent a conclusion unsupported by the source data?

Those are questions of fidelity and reliability.

Now suppose the report is offered in litigation. Can the proponent establish the foundation required for admission? Are there applicable hearsay issues? Does the report involve opinions subject to expert testimony requirements?

Those are legal questions.

And even if the report is admitted, how much weight should the factfinder assign to its conclusions? That’s yet another question.

Simply identifying the report as AI-generated doesn’t answer any of them.

Nor does identifying it as human-generated. The method of creation matters, but it isn’t a substitute for evaluating the evidence.

Different questions, different responsibilities

I think this framework also helps clarify the responsibilities of forensic experts, attorneys, courts, and factfinders.

A forensic expert investigates the evidence, evaluates technical findings, identifies limitations, and explains the strength of the resulting conclusions. Attorneys determine how those findings relate to the legal issues, evidentiary requirements, and litigation strategy. Courts decide questions of admissibility under the applicable rules. And factfinders ordinarily determine the weight and persuasive significance of admitted evidence.

These responsibilities overlap, but they aren’t interchangeable.

A forensic expert shouldn’t confuse demonstrating that a file hasn’t changed with establishing that its contents are true. An attorney shouldn’t confuse successful authentication with proof of reliability. And neither should assume that admission into evidence means the factfinder must believe it.

The future of evidence

As artificial intelligence becomes increasingly involved in creating and analyzing information, I suspect we’ll spend less time debating whether evidence was touched by AI and more time examining what that involvement means.

But the underlying questions remain familiar:

• What is the evidence?
• Where did it come from?
• What does it actually represent?
• How much confidence can we place in our conclusions?
• Can it be admitted?
• And ultimately, how much should we rely upon it?

These questions are related, but they serve different purposes.

Integrity tells us whether evidence has changed; fidelity tells us whether it faithfully represents what it claims; confidence describes how certain we are about those conclusions.

Likewise, authentication establishes what evidence purports to be, admissibility determines whether it may be considered, and evidentiary weight decides how much significance it deserves.

And trustworthiness connects those inquiries to the fundamental question at the heart of every legal dispute: What should we believe, and why?

Closing thoughts: Join the conversation

This is just one piece of the bigger conversation on the future of evidence. As legal professionals, we need to stay on top of emerging technologies.

Let’s continue the discussion on this LinkedIn post.