A Federal Judge Calls AI Deepfake Defamation Uncharted Territory

A Federal Judge Calls AI Deepfake Defamation "Uncharted Territory"

A federal court in Tennessee let a defamation claim over AI-generated videos survive dismissal on September 16, 2026, finding no caselaw that it or the parties could identify on whether AI-generated video can defame. The judge let the claim proceed to allow the record to develop rather than decide the question now.

Information last verified on September 27, 2026. This is a developing story; we update it as the record changes.

Jurisdiction scope: This article addresses Tennessee defamation law as applied by a federal district court sitting in diversity in the Eastern District of Tennessee. It does not address any state's criminal deepfake or digital-likeness statutes, which operate separately from a defamation claim. For those, see our state-by-state deepfake law tracker.

What Happened

According to the operative complaint as summarized by the Court, Randolph Duane Ross is a track and field athlete who has held collegiate coaching positions for about ten years, including Head Coach and Director of Track and Field at the University of Tennessee beginning in 2022. Greg Lambert, a former collegiate runner, approached Ross seeking to join his staff, and Ross hired him in August 2022 as Track and Field Recruiting Coordinator.

Ross alleges Lambert was unreliable from the outset and that in May 2023 Ross told him he was being terminated, offering the option to resign in lieu of being fired for cause, which Lambert accepted.

What followed is the subject of the suit. Ross alleges Lambert filed an EEOC complaint that closed without a probable cause finding against either Ross or the university. Then, around May 30, 2024, Lambert sent a demand letter to the university's Office of General Counsel threatening to sue for wrongful termination and discrimination unless he received a $150,000 payment, and including allegations that Ross and Ross's son were involved in "doping." The letter warned he would go to the media if the university did not settle. The university rejected the demand, calling his attack on Ross's reputation "baseless" and "highly offensive."

Ross alleges Lambert never filed suit and instead began what the complaint calls a "vindicative and wide-ranging campaign" across the track and field community, including, between December 12 and 25, 2025, more than 20 videos on Instagram and Facebook. Among them, per the complaint, were a fake sexually suggestive video depicting Ross with Sean "Diddy" Combs, and another AI-generated video depicting Ross in prison with Combs and Sherrone Moore, whom the opinion describes as "a disgraced former university football coach."

Lambert moved to dismiss under Rules 12(b)(2), 12(b)(3) and 12(b)(6). The Court denied the personal jurisdiction and venue challenges outright and granted the failure-to-state-a-claim motion only in part. Ross's complaint pleads two counts, defamation and false light invasion of privacy. No count was dismissed in full; the Court's Rule 12(b)(6) analysis narrowed the defamation count allegation by allegation.

The passage that matters

The Court worked through the defamation claim allegation by allegation. On the AI videos it stopped:

"Unsurprisingly given the nascent nature of AI, the Court cannot find, nor do the parties identify, any case law on defamation in relation to AI content, and specifically, videos which are AI-generated." Ross v. Lambert, No. 3:26-CV-13-TAV-JEM, slip op. at 35 (E.D. Tenn. Sept. 16, 2026)

It then restated the governing Tennessee standard, that an allegedly defamatory statement must be "judged within the context it is made" and "should be read as a person of ordinary intelligence would understand [it] in light of the surrounding circumstances," and reached its conclusion:

"In attempting to evaluate defamatory statements in accordance with these principles, AI content leads the Court into uncharted territory. Considering this, the Court finds it most appropriate to allow plaintiff's defamation claim as to these AI-generated videos to continue on at this stage of the litigation to allow further record development." Ross, slip op. at 36

The Court cited Clark v. E! Entertainment Television, LLC, 60 F. Supp. 3d 838, 848 (M.D. Tenn. 2014), which the opinion renders as "Clark v. EA Ent. Television", a Middle District of Tennessee case noting the Sixth Circuit has agreed there are "difficulties inherent in establishing defamation in cases involving television programs" because of varying segments of audio and video, editing and tone. The analogy is doing the work: a synthetic video, like an edited broadcast, is not a sentence you can read for its plain meaning.

What did not survive

This was not a wholesale win. The Court dismissed the defamation claim as to a video that "does readily highlight the UT track and field program" but where the plaintiff "is not sufficiently connected to any of the statements made during the video." It also dismissed the claim as to two songs, titled "3-years-left" and "kinda-nice," which it found "too vague to be actionable."

The Court summarized: "Accordingly, for all the reasons set forth above, most, but not all, of plaintiff's defamation allegations will proceed."

Truth is a defense, but not yet

Lambert argued Ross had not "established falsity as a matter of law" and that publicly available records showed his statements were true or at least not verifiably false. The Court agreed with the premise and rejected the timing:

"It may very well be that none of [d]efendant's statements at issue in this case are false, and that therefore[,] [p]laintiff[’s] defamation claim will fail. But undertaking to determine the truth of [d]efendant's statements at this stage would contravene the entire legal framework surrounding motions to dismiss." Ross, slip op. at 37 (quoting Church of God in Christ, Inc., 2026 WL 2256336, at *7)

On a Rule 12(b)(6) motion the court construes the complaint in the plaintiff's favor and accepts its allegations as true. Any truth defense is therefore premature.

The Court also declined to decide whether Ross is a limited-purpose public figure or whether the statements addressed a matter of public concern, holding the question unnecessary because Ross "has sufficiently alleged that defendant acted with malice." Separately, it restated that Tennessee does not presume defamation damages: "actual damage must be sustained and proven in all defamation cases."

What the Law Actually Says

Tennessee defamation requires a false and defamatory statement of fact about the plaintiff, published to a third party, with the requisite fault, causing injury. As the Court noted quoting West v. Media General Convergence, "truth is, almost universally, a defense." Our Tennessee defamation guide sets out the state's elements and its split limitation periods: one year for libel under Tenn. Code Ann. 28-3-104(a)(1)(A), but only six months for slander under Tenn. Code Ann. 28-3-103.

Three familiar requirements explain how the court sorted the allegations, though the opinion does not set them out as a formal three-part test and we group them here for clarity. They are the same three that will sort most deepfake defamation claims.

The first is that the statement must be about the plaintiff. That is what defeated the UT program video: the video highlighted the program, but Ross was not sufficiently connected to the statements in it. A synthetic video that depicts a recognizable person clears this easily, which is precisely why deepfakes are a defamation problem rather than a general-falsehood problem.

The second is that the statement must be capable of a defamatory meaning, judged in context as a person of ordinary intelligence would understand it. This is the requirement the AI videos put under strain. The context of a labeled, obviously synthetic video arguably tells the ordinary viewer that nothing depicted happened. The context of an unlabeled one tells the viewer the opposite. Most real cases sit between those poles, with a disclaimer somewhere in a caption and a video that looks real.

The third, which the court did not need to reach separately here, is that a statement must assert or imply a fact rather than opinion. A video does not "assert" in the way a sentence does; it depicts. Whether a depiction can be a factual assertion at all, and whether labeling it as artificial defeats that, is part of the open question the court identified.

This is separate from the growing body of state statutes on synthetic media. Those laws generally create their own causes of action or criminal offenses for specific categories such as election content, sexual imagery and digital replicas of performers, and they do not turn on defamation elements. Tennessee's own provisions are covered in our Tennessee deepfake law page, and the national picture is in our deepfake law tracker. We recently covered a different corner of the same problem in a suit over an AI search summary, where the alleged publisher was the AI system itself rather than a person using one.

Analysis: Why This Matters

The following is analysis from the Recording Law Editorial Team.

The temptation is to read this opinion as a court saying deepfakes can be defamatory. It does not say that. It says nobody has decided the question, that the existing test does not obviously fit, and that the claim should go forward so there is a record to decide it on. That is a modest ruling, and its modesty is the most interesting thing about it.

Consider what a court would need in order to apply Tennessee's "person of ordinary intelligence" standard to a synthetic video. It would need to know how the video was presented, what accompanied it, whether the caption disclosed the artificial origin, how prominently, whether viewers in fact understood it as fabricated, and whether the platform labeled it. None of that is in a complaint. All of it is in discovery. A judge who dismissed on the pleadings because the defendant said the content was labeled as artificial would be deciding an empirical question about audience understanding from the face of a complaint, which is the one thing Rule 12(b)(6) forbids.

So the procedural posture and the doctrinal novelty point the same way, and that is likely to be the pattern. Expect early deepfake defamation cases to survive dismissal not because courts have concluded synthetic video is actionable, but because the fitness of that ordinary-intelligence standard cannot be assessed without facts. The meaningful rulings are likely to come at summary judgment.

The second point worth drawing out is the analogy the Court chose. It reached for a television case about editing, audio and tone rather than for anything about technology. That is a sensible instinct. Defamation law has handled manipulated media before, and the doctrinal problem posed by a deepfake is a sharper version of the problem posed by a selectively edited broadcast or a doctored photograph. Six days later, in Posey v. Bushnell, No. 52072 (Idaho Sept. 22, 2026), the Idaho Supreme Court affirmed a $1,176,000 defamation judgment, $926,000 compensatory plus $250,000 punitive, in a case where the manipulation was a blur and a splice rather than a generative model. Courts do not need a new cause of action to reach synthetic media; they need to work out how the context requirement applies when the medium itself is the falsehood.

Finally, the parts of the claim the Court dismissed are as instructive as the part it kept. Two songs were too vague. A video about the program did not name the plaintiff clearly enough. Those dismissals are a reminder that the "wide-ranging campaign" framing of a complaint does not survive contact with element-by-element analysis. A plaintiff in a case like this wins or loses on specific publications, not on the pattern.

How This Affects You

If someone has published AI-generated content depicting you, this opinion suggests a defamation claim is not foreclosed by the fact that the content is synthetic, at least at the pleading stage in this district. It also suggests the analysis will be fact-intensive and will turn on how the content was presented.

If you publish AI-generated content about identifiable people, the practical reading is that a disclaimer is evidence, not immunity. The defendant here argued his content was accompanied by commentary referencing its artificial nature; the Court did not treat that as ending the inquiry on a motion to dismiss.

Courts have generally held that context governs whether a statement carries a defamatory meaning, which cuts both ways here. A clearly labeled parody sits on stronger ground than an unlabeled fabrication, but where exactly the line falls has not been decided for synthetic video by any court this opinion or the parties could find.

None of this is advice about your situation. Whether particular content is actionable, in which state, and within what deadline, are questions for a lawyer licensed where you are.

What Happens Next

The case continues in the Eastern District of Tennessee before Judge Varlan. The motion to dismiss was denied on personal jurisdiction and venue, and granted only in part on the merits, so most of the defamation allegations, including those about the AI-generated videos, remain live. The case is already well into discovery: a scheduling order issued February 6, 2026, and Lambert answered and filed a counterclaim on June 25, 2026. The substantive question the Court identified as uncharted will not be answered until a later ruling, most plausibly on summary judgment after a record exists about how the videos were presented and understood. We will update this article when the court rules again.

This is general legal information, not legal advice. It covers Tennessee defamation law as applied in Ross v. Lambert, No. 3:26-CV-13-TAV-JEM (E.D. Tenn. Sept. 16, 2026), and reflects sources verified on September 27, 2026. Allegations described here are allegations in a complaint that the court was required to accept as true at this stage; they have not been proved. Laws change and this story is developing. Consult a lawyer licensed in your jurisdiction about your specific situation.

Related articles

• Tennessee's deepfake and digital-likeness rules
• Which states regulate synthetic media, and how
• Tennessee defamation elements and deadlines
• A separate suit over an AI-written search summary

---

Last updated: 2026-09-27. This is a developing story; details verified as of 2026-09-27.

Frequently Asked Questions

Did a court just rule that deepfakes are defamation?

No. In Ross v. Lambert the court held only that the claim survives a motion to dismiss. It expressly said it could find no caselaw on defamation in relation to AI-generated video, that the area is 'uncharted territory,' and that it would allow the claim to continue 'to allow further record development.' That is a decision to postpone the question, not to answer it.

What case is this and who decided it?

Ross v. Lambert, No. 3:26-CV-13-TAV-JEM, a Memorandum Opinion and Order entered September 16, 2026 by United States District Judge Thomas A. Varlan in the Eastern District of Tennessee, ruling on the defendant's motion to dismiss.

Does labeling a video as AI-generated protect you from a defamation claim?

Not automatically, on this ruling. The defendant argued his content was accompanied by commentary referencing its artificial nature and did not attribute specific statements to the plaintiff. The court did not treat that as resolving the claim at the pleading stage, where it must accept the complaint's allegations as true.

What parts of the claim were dismissed?

The court dismissed the defamation claim as to three allegation sets: an email to UT athletics containing a 'bizarre fictional script' and links to over 50 songs, where the court found no statement capable of defamatory meaning; a fake BALCO newscast video that highlighted the University of Tennessee track and field program but to which the plaintiff was 'not sufficiently connected'; and two songs, '3-years-left' and 'kinda-nice', which it found 'too vague to be actionable.' It summarized that most, but not all, of the defamation allegations would proceed.

Why did the court refuse to consider the defendant's truth defense?

Because of the procedural stage. On a Rule 12(b)(6) motion a court construes the complaint in the light most favorable to the plaintiff and accepts its allegations as true. The court said determining the truth of the statements at that stage 'would contravene the entire legal framework surrounding motions to dismiss,' so any truth defense was premature.

How is a deepfake defamation claim different from a state deepfake statute?

They are separate routes. Defamation is a common-law tort requiring a false statement of fact about the plaintiff, published to others, with fault and injury. State synthetic-media statutes generally create their own civil or criminal liability for defined categories such as election content, non-consensual sexual imagery or unauthorized digital replicas, and do not require proof of the defamation elements.

What legal test will decide whether an AI video is defamatory?

The court applied the standard Tennessee framing, that an allegedly defamatory statement is judged within the context it is made and read as a person of ordinary intelligence would understand it in light of the surrounding circumstances. What the court said it could not do was apply that test to synthetic video without a factual record. How it applies remains undecided.

Is this ruling binding on other courts?

No. It is a district court decision on a motion to dismiss, which is persuasive rather than binding even within the Sixth Circuit. Its value is that it is among the first written analyses of the question, and it identifies why the existing test is hard to apply.