In 2007, a dispute over a lightning-struck yacht produced the most quoted opinion in the law of electronic evidence, and it did so by way of total disaster. The boat was the Chessie, anchored in the Chesapeake Bay; the fight was a roughly $22,000 insurance coverage squabble that would not ordinarily merit a footnote. But both sides moved for summary judgment, both relied almost entirely on emails, and neither had authenticated a single one of them. So the court, faced with two stacks of inadmissible exhibits, denied both motions and wrote an opinion that ran past a hundred pages explaining everything the lawyers should have done. Lorraine v. Markel American Insurance Co., 241 F.R.D. 534 (D. Md. 2007). Judge Paul W. Grimm, who wrote it, described the parties' failure with a phrase that has haunted litigators ever since: it was, he said, "a self-inflicted injury which can be avoided by thoughtful advance preparation."

That is the whole subject of this article, transposed from email to the modern problem that dwarfs it. Web content—live pages, archived pages, and social media posts—now sits inside nearly every case of any complexity. It is relevant, it is often decisive, and it is excluded with depressing regularity, not because the law is hostile to it but because counsel never lays the foundation. The authentication bar is low. The mistake of walking up to it unprepared is common. This guide is about clearing the bar deliberately, every time, and it does so by following one fictional dispute through every problem the evidence can throw at you.

A Case That Lives Entirely on the Internet

Meet Harborview Coffee Roasters (a hypothetical, like every party and fact in this guide). Harborview is convinced that its competitor, Cedar & Pine Coffee, is lying to customers. Cedar & Pine's website advertises beans that are "single-origin, direct-trade, and roasted within 48 hours of shipping"—claims Harborview believes are false—and Harborview wants to bring a false-advertising suit under Section 43(a) of the Lanham Act, 15 U.S.C. § 1125(a). (For the substantive doctrine behind that kind of claim, see our guide to false advertising and Lanham Act Section 43(a).) Every scrap of proof Harborview needs lives on the internet, and it arrives in three flavors that turn out to raise three different authentication problems.

First, Harborview must prove what Cedar & Pine's website says right now. Second, it must prove what the website said two years ago, when the allegedly false claims first appeared and Harborview began bleeding customers—except Cedar & Pine has since quietly rewritten the page, so the old version survives only in the Internet Archive's Wayback Machine. Third, Harborview has found a post on a Cedar & Pine executive's social media account that appears to admit, in an unguarded moment, that the beans actually come from a commodity broker. Each piece is potentially case-dispositive. And each can be excluded—leaving Harborview with a meritorious claim it cannot prove—if counsel fumbles the foundation. Hold those three problems in mind. We will solve them one at a time.

The Authentication Threshold Is Low—and Routinely Missed

Start with the rule everyone cites and too few satisfy. Federal Rule of Evidence 901(a) provides that, to authenticate an item, "the proponent must produce evidence sufficient to support a finding that the item is what the proponent claims it is." That standard is deliberately modest. It does not demand proof of authenticity to a certainty, or even by a preponderance; it demands only enough evidence that a reasonable jury could find the item genuine. The judge plays gatekeeper under the conditional-relevancy principle of Rule 104(b), deciding whether the foundation clears that low bar. If it does, the ultimate question of authenticity goes to the jury, which remains free to disbelieve the exhibit entirely. As the Fourth Circuit put it, the court "need not find that the evidence is necessarily what the proponent claims, but only that there is sufficient evidence that the jury ultimately might do so." United States v. Zhu, 854 F.3d 247, 257 (4th Cir. 2017). The same standard governs a yellowed paper letter and a digital screenshot alike. Lebewohl v. Heart Attack Grill LLC, 890 F. Supp. 2d 278, 298 (S.D.N.Y. 2012).

So why is a threshold this forgiving missed so often? Because electronic evidence does not explain itself the way paper does. A letter announces its own provenance through letterhead, a signature, the texture of the page, a postmark. A screenshot announces nothing. It is a rectangle of pixels that could have been captured from a live site, fabricated from scratch in an image editor, or altered after the fact with a few keystrokes in a browser's developer tools. Because the medium invites doubt, the proponent has to supply the context that a physical document carries on its face. Harborview's screenshots of Cedar & Pine's site are not admissible merely because Harborview's lawyer vouches for them; counsel must show how the captures were made and why they reflect what the site actually displayed. The good news, as we will see, is that doing so is usually a matter of preparation rather than genius.

The Rule 901(b) Toolkit

Rule 901(b) supplies a non-exhaustive menu of authentication methods—"examples only," the rule says, "not a complete list." Think of them not as a checklist to grind through but as tools to select among, matched to the evidence in hand. The leading practitioner treatise on the subject, the Practical Law practice note E-Discovery: Authenticating Electronically Stored Information—co-authored, fittingly, by Judge Grimm himself—catalogs the methods that actually work for digital evidence. A handful matter most for web content.

The most direct is testimony of a witness with knowledge under Rule 901(b)(1): someone who viewed the page and captured it testifies that the exhibit accurately reflects what appeared on the screen. The witness needs no technical expertise—only personal knowledge that they navigated to a particular URL, saw the content, captured it, and that the exhibit matches what they saw. This method carries current website evidence beautifully, which is why it will do most of the work for Harborview's live captures. It works poorly for historical evidence, because no witness can testify from personal knowledge about what a site displayed two years ago. That gap is precisely what the Wayback Machine exists to fill.

A second tool is distinctive characteristics under Rule 901(b)(4): the contents of a page can themselves supply circumstantial proof of authenticity. A corporate site bearing the company's logo, mailing address, phone number, employee names, and product catalog carries internal hallmarks that tie it to the company; a social-media profile displaying a person's photo, employer, hometown, and references to their actual life carries marks tying it to that person. Courts weigh the totality—no single feature is dispositive, but an accumulation of consistent, hard-to-fake details can satisfy the standard. This is the workhorse of website authentication. In Perfect 10, Inc. v. Cybernet Ventures, Inc., 167 F. Supp. 2d 1114 (C.D. Cal. 2001)—the foundational website-screenshot case, decided years before the Rules were updated for digital evidence—the court admitted printouts of webpages over an authentication objection, reasoning that the exhibits' own distinctive features (their content, dates, URLs, and circumstantial indicia) combined with a declaration about how they were collected were enough to satisfy Rule 901. Judge Grimm later distilled the Perfect 10 approach into a list of factors courts weigh for website printouts: how long the content stayed posted, whether others reported seeing the same content, whether the site's owner published the same content elsewhere, and whether the printout shows the URL and the date and time of access. The lesson is that a screenshot capturing more context is not just more persuasive—it is more authenticatable.

A third tool, increasingly the center of gravity for digital evidence, is evidence describing a process or system under Rule 901(b)(9): testimony or certification explaining that a system produces an accurate result. This is the doctrinal home of Wayback Machine authentication (explaining how the Internet Archive's crawlers capture and reproduce pages) and of forensic-capture-tool authentication (explaining how a tool records content, logs metadata, and generates verifying hash values). Two further tools round out the kit. Public records authentication under Rule 901(b)(7) handles government-agency sites—SEC EDGAR filings, USPTO trademark and patent records, court dockets—and pairs naturally with the self-authentication provisions for official publications in Rule 902(5). And comparison by an expert or the trier of fact under Rule 901(b)(3) lets a disputed screenshot be matched against a known-authentic specimen.

Before you authenticate the hard way, try the easy way

Crucially, before reaching for any of these tools, counsel should consider three shortcuts that can make formal authentication unnecessary—and that the best litigators reach for first. A stipulation is the cleanest path: ask opposing counsel to agree the evidence is authentic. This is especially appropriate when the evidence comes from the opponent's own website, because there is rarely a good-faith basis to dispute that your own site said what a screenshot plainly shows, and a refusal to stipulate can later make a court skeptical of a belated authenticity challenge. A request for admission under Federal Rule of Civil Procedure 36 accomplishes the same thing formally and creates a record: serve Cedar & Pine a request to admit that a specific screenshot accurately reflects its website on a specific date, and an admission—or a failure to respond within thirty days—conclusively removes the issue. (On building those requests, see our guide to strategic responses to requests for admission.) Judicial notice under Rule 201 can dispatch indisputable facts from authoritative sources—but, as we are about to see, it is a trap when aimed at the Wayback Machine. There is also a fourth, quieter route worth knowing: authentication by production. When a party produces a document from its own files in discovery, many courts treat that act as an implicit concession that the document is what it appears to be, foreclosing a later authenticity objection by the producing party. If Cedar & Pine produces a screenshot of its own historical page, it has likely authenticated it for Harborview.

The Lorraine Framework: Five Hurdles, Not One

Judge Grimm's enduring contribution in Lorraine was to refuse to treat admissibility as a single question. Electronic evidence, he insisted, must clear five hurdles, each of which the proponent must consciously address before the exhibit reaches the jury. The evidence must be relevant (Rules 401–402). It must be authenticated (Rule 901). If it contains out-of-court statements offered for their truth, it must survive hearsay (Rules 801–807). It must satisfy the original writing (best evidence) rule (Rules 1001–1008). And its probative value must not be substantially outweighed by the dangers of unfair prejudice, confusion, or waste of time under Rule 403.

These five rules govern all evidence, paper and pixel alike. What makes electronic evidence distinctive is that it presents a fresh problem at every stage, because a digital file can be altered, spoofed, fabricated, or stripped of context in ways a physical document cannot. The takeaway for Harborview's counsel is sobering and clarifying at once: authenticating the screenshots is necessary but not sufficient. A perfectly authenticated screenshot of Cedar & Pine's "48 hours" claim may still draw a hearsay objection; a perfectly authenticated capture may still be excluded under Rule 403 if it is needlessly inflammatory; and the original-writing rule lurks behind every printout. We address each hurdle below, but the structural point is the one to internalize: clearing one hurdle is not winning the race. (For a deep treatment of the last hurdle, see our guide to Federal Rule of Evidence 403 and unfair prejudice.)

Problem One: Proving the Live Site

Harborview's first task—proving what Cedar & Pine's website says today—is the easiest, because counsel can simply capture the live site and put on the person who did it. The testimony establishes four unremarkable facts: the witness navigated to a specific URL, viewed the content, captured it, and the exhibit accurately reflects what they saw. If the witness made the capture personally, that ordinarily clears the Rule 901 threshold without breaking a sweat. Courts routinely admit such captures on a witness's say-so. See, e.g., Specht v. Google Inc., 758 F. Supp. 2d 570, 580 (N.D. Ill. 2010) (treating webpage printouts as authenticated where a witness with knowledge testified that they accurately depicted the sites).

But "ordinarily" is not "always," and a handful of practices make current-website evidence dramatically more durable against attack—practices Harborview should follow precisely because Cedar & Pine has every incentive to attack the proof. They cost almost nothing and convert a vulnerable exhibit into an armored one.

Capture the full page, not a cropped excerpt. A screenshot that shows the URL in the address bar, any on-page timestamps, the site's navigation chrome, and the surrounding context is far harder to attack than a lone paragraph floating without provenance. The cropped excerpt invites the question "where did this come from?"; the full capture answers it on its face. This is the Perfect 10 lesson in operational form: the more authenticating context the exhibit carries, the lower the foundation you need to build around it.

Preserve the metadata. Capture date, time, capturing device, and resolution are all probative of when and how the image was made—and all are silently destroyed the moment a screenshot is pasted into a word processor and saved as a flat picture. Keep the original file, in its native format, with its metadata intact, and produce that.

Generate a cryptographic hash at the moment of capture. A hash function such as SHA-256 reduces a file to a fixed string of characters that serves as a digital fingerprint. Change a single pixel and the fingerprint changes completely. Recording the hash at capture means counsel can later prove, mathematically, that the trial exhibit is bit-for-bit identical to what was captured—a powerful answer to any "you altered this" challenge and an increasingly expected practice in serious litigation.

For anything likely to be contested, use a forensic capture tool. Commercial services such as Page Vault, Hunchly, and WebPreserver—and open-source options built on browser-automation frameworks like Playwright—automatically record the URL, timestamp, full rendered content, metadata, and hash, and many generate a signed certificate of capture suitable for self-authentication. A do-it-yourself screenshot is fine for a routine, uncontested page; a forensic capture is the right tool when you expect a fight.

Where time allows, capture on multiple dates. Repeated captures of the same page over days or weeks strengthen the inference that the content is a stable, genuine feature of the site rather than a momentary glitch or a manipulation staged for litigation.

Self-authentication under Rules 902(13) and 902(14)

The December 2017 amendments to the Federal Rules added two provisions that quietly revolutionized the economics of digital evidence. Before 2017, authenticating electronic records usually meant calling a live witness—a records custodian, an IT technician, a forensic examiner—to the stand for every batch of exhibits. Rule 902(13) eliminated that necessity for a large class of evidence. It permits self-authentication of "a record generated by an electronic process or system that produces an accurate result," shown by a written certification from a qualified person, provided the proponent gives adverse parties advance written notice and a chance to inspect the record and the certification. (The notice procedure is borrowed from the business-records provisions of Rules 902(11)–(12).)

For Harborview, Rule 902(13) means a forensic technician's written certification—describing how the capture was made, the system used, and the hash values generated—can substitute for putting that technician on the witness stand at all. In a case involving dozens of captured pages, that single move can save days of trial time and a small fortune in expert fees. Rule 902(14) addresses the adjacent situation: data copied from an electronic device, storage medium, or file, self-authenticated by a certification (supported by a hash or other digital-identification process) that the copy is an accurate reproduction of the original. It matters most when web evidence is pulled from a forensic image of a hard drive or phone rather than captured live from the internet—for instance, a deleted page recovered from a defendant's own server. Together, 902(13) and (14) let counsel authenticate volumes of electronic evidence on paper, in advance, without a single witness—so long as the certification is solid and the notice is timely. Sloppy notice is the usual way lawyers forfeit the advantage.

Problem Two: Proving the Past with the Wayback Machine

Harborview's second task is genuinely harder. The allegedly false claims first surfaced two years ago, but Cedar & Pine has since rewritten the page, so the original survives only in the Internet Archive's Wayback Machine—a digital library run by a San Francisco nonprofit that has been crawling and storing web pages since 1996 and now holds well over 800 billion of them. The Wayback Machine works by sending automated crawlers to visit sites and store snapshots of what they find. Coverage is uneven and unpredictable: a major news homepage may be captured many times a day, a small business site only sporadically or never. Each snapshot is stamped with the date and time of capture, encoded directly into the archived URL (the long string beginning web.archive.org/web/ followed by a fourteen-digit timestamp). That timestamp is a built-in piece of metadata counsel should record and preserve for every snapshot they intend to use.

The evidentiary value is obvious and often irreplaceable. When a party must prove what a site displayed before anyone thought to capture it, the Wayback Machine may be the only source on earth. That makes it invaluable in intellectual property cases turning on priority of use or the existence of prior art, in contract cases where the terms posted at the time of a deal govern, in trademark disputes over when a mark was first used in commerce, and in exactly Harborview's situation, where the question is what a competitor claimed and when. But the authentication challenge is equally real, for the same reason that doomed live-site testimony in reverse: no human can testify from personal knowledge about what the site displayed two years ago. The proponent must instead establish that the Archive's systems reliably captured the page and faithfully reproduced it—a classic Rule 901(b)(9) "process or system" showing.

What Weinhoffer teaches: you cannot just attach it

The single most important case for any practitioner relying on the Wayback Machine is Weinhoffer v. Davie Shoring, Inc., 23 F.4th 579 (5th Cir. 2022). There, a party tried to prove the terms of a historical auction using a Wayback Machine printout, and the district court took judicial notice of it under Rule 201, treating the Internet Archive as a source "whose accuracy cannot reasonably be questioned." The Fifth Circuit reversed. Its reasoning is the holding worth memorizing: sister circuits had admitted archived snapshots only when they were properly authenticated—through testimony or an affidavit from someone with knowledge of the Archive's processes—and that very practice implied the snapshots are not so self-evidently reliable that a court may simply notice them under Rule 201's exacting "cannot reasonably be questioned" standard. Because the proponent in Weinhoffer had offered no authentication whatsoever, the printout should never have come in.

Read carefully, Weinhoffer is not a defeat for Wayback evidence. It draws a clean line between two doors. The judicial-notice door (Rule 201) is the wrong one, because it demands indisputability that a web archive cannot supply. The authentication door (Rule 901) is the right one, and it is wide open, because it demands only a prima facie showing. The practical command is unmistakable: attaching a Wayback printout to a brief and gesturing at it is malpractice waiting to happen. You must walk through the 901 door.

Earlier cases mark the same path from the other direction. Telewizja Polska USA, Inc. v. Echostar Satellite Corp., 2004 WL 2367740 (N.D. Ill. Oct. 15, 2004), is the early landmark: the court accepted an affidavit from an Internet Archive employee as sufficient under Rule 901 to authenticate archived pages, and rejected hearsay and best-evidence objections to boot. And the Second Circuit blessed the approach at the appellate level in United States v. Gasperini, 894 F.3d 482 (2d Cir. 2018), upholding admission where the government called the Internet Archive's office manager to explain how the Wayback Machine crawls, stores, and reproduces pages and how reliable the process is, with the defendant free to cross-examine on every point. Together, these cases set the floor (you must authenticate) while leaving the exact quantum to flex with the court and the vigor of the opponent's challenge. A growing line of district courts now treats a proper Internet Archive affidavit as routinely sufficient. See, e.g., Marten Transp., Ltd. v. PlattForm Advertising, Inc., 184 F. Supp. 3d 1006 (D. Kan. 2016).

How Harborview should authenticate its historical page

There are four routes, and counsel should pick by stakes and timing.

The most reliable is an affidavit from the Internet Archive itself. The Archive maintains a standardized litigation-support process (described at archive.org/legal): for a fee, it provides printed copies of specified archived pages together with a sworn affidavit from a staff member attesting to the Archive's technical processes and confirming that the printouts are true and accurate copies of what its systems captured on the stated dates. This affidavit has satisfied Rule 901 in dozens of courts and is the gold standard for contested historical web evidence. The catch is logistics: counsel must identify each page by its specific timestamped Wayback URL, and turnaround takes weeks, so a request fired off the week before a discovery deadline is asking for heartbreak. Plan the request the way you would plan a third-party subpoena—early.

A second route is live or declaration testimony under Rule 901(b)(9) from a witness who genuinely understands the Archive's operations and can explain how it captures and stores pages, how counsel retrieved the specific snapshot, and that the printout accurately reflects it. This is the Gasperini model, and it suits smaller matters where the cost and lead time of a formal Archive affidavit are not justified.

A third is a Rule 902(13) certification where counsel used a forensic capture tool to retrieve and verify the archived content—folding the Wayback snapshot into the same self-authentication framework used for live captures, complete with hash verification.

A fourth, available because the historical content is Cedar & Pine's own, is a stipulation or request for admission. Cedar & Pine has unique knowledge of what its site said two years ago. An unreasonable refusal to admit genuinely accurate content can later invite cost-shifting under Rule 37(c)(2) or an adverse inference, which gives Harborview real leverage to take the authentication question off the table entirely.

Problem Three: Proving Authorship on Social Media

Harborview's third piece—the executive's apparent admission on social media—raises a problem the first two do not: attribution. Anyone can create an account in anyone else's name, and the bare fact that a post appears on an account labeled "Jordan Cedar" proves only that an account by that name published it, not that the flesh-and-blood Jordan Cedar wrote a word of it. This is the central, recurring difficulty of social-media evidence, and it is distinct from the question of whether the screenshot accurately depicts the post.

Social-media evidence is also more varied and more fragile than a static webpage, which compounds the trouble. The category sprawls across Facebook posts, Instagram photos, posts on X, TikTok videos, LinkedIn profiles, Reddit threads, and Snapchat captures—each with its own format, metadata conventions, and retention behavior, and some of it engineered to vanish. A Snapchat image or an Instagram "story" may evaporate in twenty-four hours; a post can be deleted the instant its author smells litigation; an entire account can be deactivated overnight. This ephemerality sharpens the preservation imperative beyond anything ordinary web content requires: the executive's helpful admission may be gone before a subpoena can issue, so counsel should capture it the instant it is found—full metadata, full URL, full surrounding context—and immediately send preservation demands to both the author and the platform. A screenshot that captures the post's text but loses the account handle, the timestamp, and the URL throws away exactly the corroboration that attribution will later demand. (Preservation obligations and litigation holds for this kind of evidence are part of the broader discovery picture covered in our practical discovery refresher.)

Three camps on how hard authentication must be

Courts have split into roughly three camps on how much scrutiny a social-media post requires before it reaches the jury, and counsel must know which camp the governing court occupies, because it dictates how much corroboration to assemble.

The first camp applies the ordinary Rule 901(a) threshold and nothing more: testimony that a witness saw the post on the account, plus identifying details on the account itself, gets the exhibit in, and the question of who really authored it is left to the jury as a matter of weight. The second camp applies heightened scrutiny, alert to how trivially accounts are faked, hacked, or hijacked. The landmark is Griffin v. State, 419 Md. 343 (2011), which held that a MySpace printout was not authenticated merely by the account's display of the purported author's birthdate and photograph; the court demanded evidence "substantially eliminating" the possibility that someone else created the post—for instance, the testimony of the purported author, a search of their device, or information from the platform itself. The third camp charts a middle course, exemplified by Tienda v. State, 358 S.W.3d 633 (Tex. Crim. App. 2012): the court examines the totality of the circumstantial evidence—profile photos, a linked email address, references to the person by a private nickname, content that only that person would plausibly know—and asks whether a reasonable juror could find the post authentic. Tienda admitted MySpace pages on exactly that aggregated showing.

Whatever the local standard, the Second Circuit's caution in United States v. Vayner, 769 F.3d 125 (2d Cir. 2014), frames the task with unusual clarity. There the court reversed the admission of a printout of a Russian social-networking profile that bore the defendant's name, photograph, and accurate biographical details. The government's witness could confirm the page came from that profile, but nothing tied the profile's creation or authorship to the defendant himself—anyone, the court noted, could have made a page with that information. Proving that content came from a particular account, Vayner holds, is simply not the same as proving that a particular person authored it. For Harborview, the moral is direct: a bare screenshot of the executive's post is the weakest possible version of the evidence, vulnerable in every camp and fatal in Griffin and Vayner jurisdictions.

The strongest version is built in three layers. First, where possible, obtain platform-produced records by subpoena—account registration data, login history with IP addresses, and content records with native metadata—which carry far more weight than any third-party screenshot because they connect the account to a real-world person and device. (Civil subpoenas for content held by the platform run into the Stored Communications Act, so the practical route is usually to subpoena the user's own production or to seek non-content records; the platform-records and takedown mechanics that accompany this discovery are treated in our guide to how to file a DMCA takedown notice and respond to one.) Second, capture the full post with metadata and URL rather than a cropped image. Third, and often decisively, use the author's own testimony: depose the executive and ask, on the record, whether he wrote the post. If he confirms it, the post is authenticated by an opposing party's own admission; if he denies it, counsel has manufactured impeachment material and a road map for the platform subpoena. The deposition is the single most efficient tool here, which is one more reason to master it (see our guide to the art and science of depositions in federal civil litigation).

Authentication Is Not the Finish Line: The Hearsay Problem

Here is the trap that catches lawyers who think the work ends the moment an exhibit is admitted "for authenticity": a perfectly authenticated screenshot can still be excluded as hearsay. Hearsay is an out-of-court statement offered to prove the truth of what it asserts (Rule 801(c)), and website content is wall-to-wall statements. The analysis turns entirely on why the content is offered.

Often it is not offered for its truth at all. If Harborview offers Cedar & Pine's "roasted within 48 hours" claim to prove the beans were not roasted within 48 hours, it is not vouching for the statement's truth—it is offering the statement as the very falsehood that constitutes the wrong. That is the non-hearsay verbal act, the operative-language move, and it is the cleanest escape from the hearsay rule. Web content offered to prove that something was published—that a trademark was in use in commerce, that a party had notice of a fact, that a representation was made to consumers, that content existed on a date as prior art, or that contractual terms were posted as the operative document—does not implicate hearsay at all, because the relevance lies in the fact of the statement, not its truth. Making that non-hearsay purpose explicit to the court, in a brief or a proposed limiting instruction, can dissolve a hearsay objection before it gains traction. For Harborview, the false-advertising claim itself supplies the framing: the legally operative fact is that Cedar & Pine made the representation to the market.

When web content is offered for its truth, several Rule 803 exceptions recur. The business-records exception, Rule 803(6), is the most common and the most overclaimed. It covers records made and kept in the regular course of a business by someone with knowledge, and it can reach e-commerce pricing, transaction logs, inventory data, and employee directories—but courts repeatedly caution that casual blog posts, marketing copy, and user-generated content usually flunk its requirements of regularity, contemporaneity, and trustworthiness. A company's automated price database may qualify; a puffy "About Us" page rarely does. The market-reports-and-commercial-publications exception, Rule 803(17), reaches price lists, directories, market quotations, and databases that the relevant trade or profession relies upon. The then-existing state-of-mind exception, Rule 803(3), can admit a social-media post reflecting the author's then-existing intent, plan, motive, or feeling—useful when the relevance is the speaker's mental state rather than an external fact. And the residual exception, Rule 807, is a narrow, sparingly invoked safety valve for evidence that is demonstrably trustworthy yet fits no enumerated exception, available only on pretrial notice.

There is also a shortcut Harborview can fire straight at Cedar & Pine: a statement offered against the party that made it is not hearsay at all under Rule 801(d)(2). That is exactly how the executive's social-media admission comes in once authenticated as genuinely his—as an opposing party's statement (or, if made within the scope of his employment and during the relationship, a vicarious admission under Rule 801(d)(2)(D)). Note the dependency: the hearsay shortcut works only if authentication succeeds, which loops back to the Vayner attribution problem. Authentication and hearsay are not independent checkboxes; they interlock.

The double-hearsay wrinkle in archived pages

Wayback Machine evidence adds one more layer. An archived page may contain statements that appeared on the original site (layer one) which were then captured and reproduced by the Archive's systems (layer two). Under Rule 805, hearsay within hearsay is admissible only if each layer independently satisfies an exception. Courts have generally handled the second layer by treating the Archive's systematic, automated captures as reliable mechanical reproductions rather than human assertions—or as business records—while the original content must still clear its own hearsay hurdle at the first layer (or be offered for a non-hearsay purpose). Where the accuracy of the archived reproduction is not genuinely disputed, courts have admitted the evidence without belaboring the point, as Telewizja Polska did when it rejected the hearsay objection to Internet Archive pages. The cleanest move, again, is to identify a non-hearsay purpose at layer one in the first place—then the Rule 805 stack collapses to a single, manageable question about the Archive's reliability, which the affidavit already answers.

The Best Evidence Rule and the Other Side of the Table

Two further points complete the picture, and both are easy to overlook until they bite.

The first is the best evidence rule, Rules 1001–1008—the fourth of Judge Grimm's hurdles, the one that trips fewer lawyers than hearsay but occasionally proves quietly decisive. The rule requires an "original" to prove the content of a writing, recording, or photograph. For paper that can be a real problem; for electronically stored information the Rules define "original" so generously that the objection usually evaporates. Under Rule 1001(d), for electronically stored information, "any printout—or other output readable by sight—if it accurately reflects the information" is an original. A screenshot or printout of a webpage therefore qualifies as an original so long as it accurately reflects what the page displayed—which loops directly back to the authentication showing. Rule 1003 adds that a duplicate is admissible to the same extent as an original unless there is a genuine question about the original's authenticity or it would be unfair to admit the duplicate in the original's place. The practical effect for Harborview is reassuring: its accurate, well-captured screenshots and printouts are "originals" in the eyes of the Rules, and a best-evidence objection has teeth only when the capture is sloppy enough to raise a real question whether the printout faithfully reflects the page. That is one more dividend of the full-page, metadata-preserving, hash-verified capture: it forecloses the original-writing objection along with the authentication one.

The second point is that you will stand on both sides of this evidence over a career, and the same framework that helps Harborview offer web evidence helps a defendant attack it. When Cedar & Pine receives Harborview's screenshots, its counsel should interrogate precisely the things this guide tells a proponent to shore up. Is the URL visible, or could the content have come from anywhere? Is there metadata, or was it stripped when the image was pasted into a brief? Is there a hash establishing the file is unaltered since capture? For the Wayback evidence, did Harborview authenticate it through an affidavit or testimony, or did it merely staple a printout to a motion in defiance of Weinhoffer? For the social-media post, did Harborview prove the executive authored it, or only that an account in his name published it—the Vayner gap? A defendant who frames an authenticity challenge around concrete, identified deficiencies stands a real chance of keeping damaging evidence out. A defendant who simply intones "objection, foundation" without naming what is missing usually loses, because the threshold is low and courts have little patience for reflexive, tactical obstruction. The asymmetry rewards specificity on both sides. Counsel who masters this toolkit wins as a proponent and as an opponent; the tools are the same, only the direction of fire changes.

The Deepfake Era: Rule 707 and Proposed Rule 901(c)

Looming over all of this is a development that threatens the foundational assumption of visual and auditory evidence: generative AI can now produce synthetic images, audio, and video that are difficult to distinguish from genuine recordings. Fabricating a screenshot was never hard—a browser's developer tools and a text editor have long sufficed to make a webpage "say" anything—but AI has raised the salience of fabrication across every medium at once, made authenticity challenges more frequent, and handed every opponent a ready-made talking point. Courts have already begun to wrestle with it: in one widely noted criminal matter, defense counsel objected that AI-driven image processing on a tablet might have altered zoomed-in video, and in others, judges have confronted "it's a deepfake" objections to recordings with visible uncertainty about the governing standard.

The federal rulemakers have responded, and practitioners need the current status rather than last year's gossip. The centerpiece is Proposed Federal Rule of Evidence 707, captioned "Machine-Generated Evidence." It addresses a specific gap: a party can sometimes offer the output of a machine or software process—an algorithm's conclusion, a model's result—without calling any human expert to sponsor it, thereby slipping past the reliability gatekeeping of Rule 702 and Daubert. Proposed Rule 707 would close that gap by providing that when machine-generated evidence is offered without an expert witness, and the evidence would be subject to Rule 702 if a human had testified to the same conclusion, the court may admit it only if it satisfies the reliability requirements of Rule 702(a)–(d). In short: the output of a black-box process gets the same Daubert scrutiny a human expert would. The proposal exempts the routine output of basic scientific instruments. Its catalyst was the now-infamous Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023), where lawyers filed a brief studded with case citations that ChatGPT had simply invented. The Advisory Committee on Evidence Rules voted to publish Rule 707 for public comment in 2025; the comment period closed on February 16, 2026, and the Committee is expected to take up the comments and issue a report in mid-2026. On the conventional Rules Enabling Act timetable, a surviving rule would then move through the Judicial Conference, transmission to the Supreme Court by roughly May 1 of the following year, and—absent congressional intervention—a default effective date of December 1, 2027. Comments flagged real difficulties, including that the operative term "machine-generated evidence" is not crisply defined and that the rule may be hard to administer precisely when no expert is available to be examined, so the final text may yet be refined.

A companion proposal, Rule 901(c), would address deepfakes head-on through a burden-shifting two-step: a party challenging an item as AI-fabricated would first have to come forward with enough evidence to support a finding of fabrication—a bare accusation would not suffice—and only if that threshold were met would the proponent have to authenticate under a standard more demanding than the usual prima facie showing. The design is deliberate: it guards against the "deepfake defense" being weaponized to manufacture doubt about plainly authentic evidence, while still giving courts a tool when fabrication is genuinely in play. As of this writing the Advisory Committee has concluded that a deepfake-specific amendment is not necessary at present—courtroom deepfakes remain rare, and existing tools (Rule 403, the judge's gatekeeping role, and expert testimony) may suffice—but it has kept Rule 901(c) on its agenda for future consideration rather than abandoning it. Meanwhile the states have moved: Louisiana enacted the first deepfake-evidence statute, Act No. 178 (effective August 2025), directing attorneys to exercise reasonable diligence to verify the authenticity of evidence with attention to AI fabrication, and the National Center for State Courts has circulated guidance and bench cards to help trial judges manage the issue.

For Harborview, the deepfake era only sharpens a lesson the authentication framework has been teaching for decades: the stronger the foundation, the harder the attack. A screenshot captured with a forensic tool, accompanied by metadata, a SHA-256 hash, and a technician's Rule 902(13) certification, is far more resistant to a fabrication challenge than a bare image with no provenance—because the proponent can prove, mathematically and procedurally, exactly when and how the file came to exist. As synthetic evidence grows more plausible, rigorous capture practices migrate from "best practice" toward "necessity." These developments sit within the broader landscape of AI in litigation, which we survey in artificial intelligence key legal issues.

From Capture to Courtroom: The Practical Sequence

Pull the threads together and the work of getting web evidence admitted has a natural order. Following it is what separates the practitioner who wins on this evidence from the one who watches it get struck.

It begins long before trial—often on the day the case walks in the door. As soon as relevant web content is identified, capture and preserve it immediately, because websites change and posts vanish, using the best available method (a forensic tool for anything likely to be contested, a metadata-preserving native screenshot at the floor). Make sure any litigation-hold notice expressly covers websites and social media, not just email and documents; the duty to preserve attaches the moment litigation is reasonably anticipated, and spoliation of web evidence draws the same sanctions under Rule 37(e) as any other ESI. For historical content, search the Wayback Machine early and record the timestamped URL of every relevant snapshot, even if you do not yet need it—snapshots can disappear, and the Archive occasionally honors removal requests.

During discovery, work the shortcuts before the formal proofs. Request stipulations for evidence from the opponent's own site and document any refusal in writing. Serve requests for admission that specific captures accurately reflect the opponent's site on specific dates. Subpoena platform records for any contested social-media post. Request the Internet Archive affidavit well ahead of any deadline. And depose the opponent on its own web content, because a deposition admission can authenticate an exhibit and dispatch a hearsay objection in a single stroke. (These moves live inside the larger discovery toolkit covered in our guides to mastering document discovery and mastering interrogatories.)

Match the authentication method to the evidence. The table below distills the pairings, but the underlying judgment is simple: use the lightest method that will actually hold up against the challenge you realistically expect.

Evidence type Primary method Backup methods
Current website (your own capture) Witness testimony, Rule 901(b)(1) Forensic capture + Rule 902(13) certification; hash verification
Current website (third-party capture) Witness testimony + distinctive characteristics, Rule 901(b)(4) Rule 902(13) certification if a forensic tool was used
Wayback Machine (historical content) Internet Archive affidavit Rule 901(b)(9) testimony; Rule 902(13) certification; stipulation/admission
Social media (your own capture) Witness testimony + metadata + distinctive characteristics Forensic capture; platform-produced records; author's deposition
Social media (platform records) Custodian or business-records certification, Rule 902(11)/(12) Request for admission; deposition testimony
Government website Rule 901(b)(7) + Rule 902(5); judicial notice for indisputable facts Witness testimony; Rule 902(13) certification

On hearsay, decide for each item, in advance, whether it is offered for its truth or for a non-hearsay purpose (notice, the fact of publication, operative language). If for its truth, select the right exception before trial rather than improvising at the lectern—and for Wayback evidence, be ready to argue the Rule 805 double-hearsay point. Finally, in trial preparation, a motion in limine can force a pretrial ruling on admissibility, leaving time to cure any shortfall instead of discovering the gap mid-trial. Build in backup authentication—a certification standing behind a witness, a witness standing behind an affidavit—so a single failure does not sink the exhibit. And in high-stakes matters, anticipate a fabrication challenge and have the hashes, metadata, forensic certifications, and corroborating testimony staged as rebuttal, along with a proposed limiting instruction wherever evidence comes in for a narrow purpose.

Frequently Asked Questions

Is a screenshot of a webpage admissible in federal court? Yes, routinely—if it is properly authenticated and clears the other evidentiary hurdles. A screenshot is treated as an "original" under Rule 1001(d) so long as it accurately reflects what the page displayed. The recurring failure is not the screenshot itself but the missing foundation: counsel must show, through a witness, a certification, or circumstantial indicia, that the exhibit is what it purports to be.

Do I need an expert to authenticate a website screenshot? Usually not. For a live page you captured yourself, your own testimony that you visited the URL, saw the content, and captured it accurately is ordinarily enough under Rule 901(b)(1). Experts and forensic tools become valuable when the evidence is contested, when you want to self-authenticate under Rule 902(13) without live testimony, or when you anticipate a fabrication challenge.

Can I just take judicial notice of a Wayback Machine page? No—that is the precise mistake Weinhoffer v. Davie Shoring condemned. Judicial notice under Rule 201 requires a source whose accuracy "cannot reasonably be questioned," and a web archive does not meet that bar. The correct route is authentication under Rule 901, most reliably through an affidavit from the Internet Archive or testimony about how the Wayback Machine works.

How do I get an Internet Archive affidavit, and how long does it take? The Internet Archive offers a standardized litigation-support service (see archive.org/legal). You identify the pages by their specific timestamped Wayback URLs, pay a fee, and receive certified printouts with a staff affidavit attesting to the Archive's processes. Turnaround runs weeks, so request it early—treat it like a third-party subpoena, not a last-minute errand.

A post on the defendant's social media account hurts their case. Is a screenshot of it enough? It is the weakest version of the evidence. A screenshot proves an account published the content, not that a particular person authored it—the gap the Second Circuit flagged in United States v. Vayner. In a heightened-scrutiny jurisdiction following Griffin v. State, a bare screenshot will likely fail. Strengthen it with platform-produced records, full metadata and the URL, distinctive content only that person would know, and, best of all, the author's own deposition testimony.

Is website content hearsay? It depends entirely on why you offer it. Content offered to prove that a statement was made—notice, the fact of publication, a false advertising claim, operative contract terms—is not hearsay at all. Content offered for the truth of what it asserts is hearsay unless an exception (business records, market reports, state of mind, residual) or an exclusion (opposing party's statement under Rule 801(d)(2)) applies.

What is a hash value and why should I bother? A hash (such as SHA-256) is a digital fingerprint computed from a file. Any change to the file, even a single pixel, produces a completely different hash. Recording the hash at the moment of capture lets you prove later that your trial exhibit is bit-for-bit identical to what you captured—an increasingly expected safeguard and a strong answer to any claim that the evidence was altered or fabricated.

Does the deepfake problem change how I should capture evidence today? Yes, at the margin. Proposed Rules 707 and 901(c) signal that courts are tightening the reliability screws on machine-generated and potentially fabricated evidence. The defensive response is the same discipline good litigators already practice: capture with a forensic tool, preserve metadata, record a hash, and obtain a Rule 902(13) certification, so the provenance of your exhibit is documented and verifiable before anyone cries "deepfake."

Conclusion

The internet now sits inside nearly every case of any complexity, and Harborview's three problems—the live page, the archived page, and the social-media post—are the problems every modern litigator meets. None of them is insurmountable. Rule 901's threshold is low. The Lorraine framework maps the five hurdles. Perfect 10 and the distinctive-characteristics cases show how a well-built capture authenticates itself. Weinhoffer tells you exactly what the Wayback Machine requires, and Gasperini and Telewizja Polska show you how to give it to a court. Griffin, Tienda, and Vayner tell you to bridge the gap from account to author. And the hearsay rules, once understood, are navigable—often dissolvable by naming a non-hearsay purpose. The federal response to AI-generated evidence—Rule 707, now past its comment period and on track toward a possible December 2027 effective date, and the still-pending Rule 901(c)—is real, but it rewards the same discipline good practitioners have always shown: capture rigorously, preserve provenance, and lay the foundation before you need it.

The lightning that struck the Chessie was an act of nature. The evidentiary failure that followed was entirely human and entirely avoidable—"a self-inflicted injury," in Judge Grimm's words, "which can be avoided by thoughtful advance preparation." The tools to avoid it are sitting right there, on your screen and in the Rules. The only open question is whether counsel will pick them up before trial or wish they had after.

For assistance with electronic-evidence strategy, preservation, capture, and authentication in federal litigation, contact our litigation practice.

Related Articles

Selected Authorities

Lorraine v. Markel American Insurance Co., 241 F.R.D. 534 (D. Md. 2007); Perfect 10, Inc. v. Cybernet Ventures, Inc., 167 F. Supp. 2d 1114 (C.D. Cal. 2001); Weinhoffer v. Davie Shoring, Inc., 23 F.4th 579 (5th Cir. 2022); United States v. Gasperini, 894 F.3d 482 (2d Cir. 2018); Telewizja Polska USA, Inc. v. Echostar Satellite Corp., 2004 WL 2367740 (N.D. Ill. Oct. 15, 2004); Marten Transp., Ltd. v. PlattForm Advertising, Inc., 184 F. Supp. 3d 1006 (D. Kan. 2016); Specht v. Google Inc., 758 F. Supp. 2d 570 (N.D. Ill. 2010); Lebewohl v. Heart Attack Grill LLC, 890 F. Supp. 2d 278 (S.D.N.Y. 2012); United States v. Zhu, 854 F.3d 247 (4th Cir. 2017); Griffin v. State, 419 Md. 343 (2011); Tienda v. State, 358 S.W.3d 633 (Tex. Crim. App. 2012); United States v. Vayner, 769 F.3d 125 (2d Cir. 2014); Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023). Fed. R. Evid. 104, 201, 401–403, 702, 801–807, 901, 902(5), 902(11)–(14), 1001–1008; Fed. R. Civ. P. 36, 37(c)(2), 37(e); Proposed Fed. R. Evid. 707 (comment period closed Feb. 16, 2026); Proposed Fed. R. Evid. 901(c) (retained on the Advisory Committee's agenda); Louisiana Act No. 178 (2025). See also Hon. Paul W. Grimm & Gregory P. Joseph, E-Discovery: Authenticating Electronically Stored Information (Practical Law Practice Note).

This article is for general informational purposes only and does not constitute legal advice, nor does it create an attorney-client relationship. The Federal Rules of Evidence and the case law applying them vary by jurisdiction and continue to evolve, and the rulemaking described here is not yet final; consult the applicable rules and qualified counsel about your specific case.