Copyright arises the moment a work is fixed in a tangible medium — no registration, no ceremony. But the moment you have to prove that the work existed, in that exact form, on that date, the law becomes demanding. For a century, creators answered with envelopes and postmarks. Today they answer with hashes, timestamps and blockchains. The interesting question is not whether courts accept cryptographic evidence — increasingly they do — but where its limits lie.

The problem cryptography solves

A copyright dispute usually turns on two questions of evidence: what is the file, and when did it exist? The first is a question of integrity — is this the same work the creator made? The second is a question of anteriority — can the creator show they possessed it before the other side?

Physical self-mailing answered both badly. As the companion to this article explains, a sealed, postmarked envelope cannot prove what was inside it at the time of posting. It fails because the chain of custody is broken by definition: the creator had unrestricted access to tamper with the evidence.

Why the sealed envelope failed

The 2009 case of Smith v. State of New York is the clearest demonstration. Mr Smith produced two envelopes, postmarked 1995 and 1996, as proof that he authored a screenplay first. The court found the contents referred to a book published years after the alleged postmarks, noted the envelopes were never produced at trial, and drew an adverse inference against him. Physical timestamping, the court concluded, is inherently equivocal and open to opposing presumptions. A postmark dates the packaging, not the creative work inside it.

The cryptographic trust architecture

SHA-256: one file, one fingerprint

A cryptographic hash function such as SHA-256 takes an input file of any size and deterministically produces a unique 256-bit string. Its legal value rests on two mathematical properties:

  • Collision resistance. It is computationally infeasible to find two different files that produce the same hash, so a given hash uniquely identifies a given file.
  • The avalanche effect. Changing a single bit — one pixel in a photograph, one comma in a document — changes the entire digest beyond recognition.

An isolated hash therefore proves integrity, but not time: it can be computed at any moment. To establish anteriority, the digest must be bound to a trusted clock. Courts and forensic standards recognise two broad architectures for doing so — centralised public-key infrastructure and decentralised ledgers.

RFC 3161 and eIDAS-qualified timestamps

The IETF standardised trusted timestamping in RFC 3161, an architecture built on institutional trust. A creator hashes their work and submits only the hash — never the underlying file — to a central Time Stamping Authority (TSA). The TSA appends a trusted clock value traceable to Coordinated Universal Time and digitally signs the packet with an X.509 key held in a hardware security module. The resulting token mathematically guarantees that the specific fingerprint existed at the recorded instant, whenever the token is later examined.

In the European Union, when this is performed by a rigorously audited Qualified Trust Service Provider under the eIDAS Regulation, it becomes a qualified electronic timestamp. Article 41(2) gives it a statutory presumption of the accuracy of the date and time and the integrity of the data to which they are bound. The practical effect is that the burden shifts to the other side to prove the evidence is fraudulent.

OpenTimestamps and Bitcoin anchoring

Where RFC 3161 relies on institutional trust, OpenTimestamps relies on distributed consensus. Thousands of hashes are aggregated into a Merkle tree whose root is embedded in a public Bitcoin transaction. Once the network confirms the block, the timestamp becomes immutable in a way that would require rewriting the global blockchain to forge. Its defining legal advantage is independent auditability: a court, an expert or opposing counsel can verify the path from the creator's local file to the public ledger without relying on the survival or honesty of a corporate TSA.

Where cryptographic evidence is winning

The European Union: eIDAS and the Marseille ruling

Because there is no mandatory pan-European copyright register, independent temporal proof is central to EU litigation, and member states have modernised their IP offices accordingly. France's e-Soleau hashes uploads with SHA-512 and applies qualified timestamping; the Benelux i-DEPOT issues a cryptographically secure dated certificate. Neither grants rights, but both satisfy the evidentiary burden of prior possession.

The eIDAS framework goes further. Article 45k of the updated eIDAS 2.0 framework prevents national courts from denying the legal effect and admissibility of electronic ledgers solely because they are electronic or decentralised.

The most concrete application came in a 2025 ruling of the Tribunal Judiciaire de Marseille, in which the fashion house AZ Factory pursued a counterfeiter. AZ Factory had anchored SHA-256 hashes of its design files to a public blockchain. Aware that French civil law scrutinises digital evidence closely, it reinforced the technical proof with a procès-verbal de constat — a sworn report by a commissaire de justice verifying the contents, hashes and dates. The court evaluated the evidence as preuve imparfaite (imperfect proof, subject to the judge's free assessment) and found the combination of blockchain timestamping and the bailiff's audit to be a credible, unbroken chain of custody. It awarded damages and ordered the counterfeit goods destroyed. The ruling is a landmark European precedent: cryptographic hashing, properly introduced alongside traditional judicial-officer reports, satisfies the rigorous demands of a civil tribunal.

The United Kingdom: chain of custody and ISO/IEC 27037

UK copyright is unregistered, so disputes in the Intellectual Property Enterprise Court and the High Court turn entirely on the claimant's ability to document creation and provenance. UK courts treat bare digital files with suspicion precisely because they are so mutable. Admissibility is governed by the Civil Evidence Act 1995 and by digital-forensics standards — above all ISO/IEC 27037, which requires that a SHA-256 hash be generated at the moment of acquisition. A match between the collection hash and the exhibit in court mathematically confirms integrity.

UK practitioners rely on a triad — hash, timestamp, chain of custody. Notably, the UK retained key eIDAS elements after Brexit: Article 24A, inserted by Statutory Instrument 2019/89, preserves the presumption of data integrity and temporal accuracy for EU-qualified timestamps in UK proceedings. Courts elsewhere in the common-law world are following: the High Court of Ireland endorsed hash-value verification in technology-assisted review in Irish Bank Resolution Corp v. Quinn. Conversely, bare screenshots and forwarded files without hash verification are routinely excluded.

The United States: FRE 902(13) and (14)

The modern US position was shaped by Lorraine v. Markel American Insurance Co. (2007), a 101-page opinion that rejected virtually all of the electronic evidence before the court because neither side had authenticated it. Lorraine made clear that digital evidence must clear a multi-step hurdle: relevance, authentication under FRE 901, hearsay, and the original-writing rule — and satisfying FRE 901(b)(9) historically meant retaining forensic experts to testify about the system that produced the file.

The Federal Rules were amended in December 2017 to add Rules 902(13) and 902(14), which allow electronic records to be self-authenticating and avoid live expert testimony. FRE 902(14) admits data copied from a device or file if authenticated by a process of digital identification — and the Advisory Committee Notes explicitly name “hash value” as that process. A copyright holder can now submit a written certification, under penalty of perjury under 28 U.S.C. § 1746, declaring that the SHA-256 hash of the exhibit matches the hash of the original file captured at a specific time. Combined with an OpenTimestamps anchor or an RFC 3161 token, that can establish independent creation and integrity and neutralise tampering defences early in discovery.

Where it still fails

Registration is still the gate

In the United States, technically perfect proof of prior possession is legally inert without registration. Under 17 U.S.C. § 411(a), no infringement action may be instituted until the Copyright Office has acted on the application — and Fourth Estate confirmed that pending applications do not count.

The point was driven home in the protracted UAB Planner 5D v. Facebook litigation (2019–2023). Planner 5D alleged that researchers scraped more than 3,000 of its 3D models to build a dataset used to train computer-vision models. It had extensive internal metadata establishing creation and ownership, but because it had not secured final registration of the specific models before filing, the court repeatedly dismissed its copyright claims. Cryptographic hashing can authenticate evidence to prove facts, but it cannot manufacture subject-matter jurisdiction.

Brittleness: the avalanche effect meets the real world

SHA-256's strictness is also a weakness. Infringers rarely leave a file byte-identical: they compress images, crop, transcode video, apply filters. If a plaintiff presents the SHA-256 of a pristine original and the court hashes the defendant's cropped copy, the hashes will not match — even though the visual content is identical. On a purely cryptographic view, a non-match could falsely suggest no infringement.

Perceptual hashing (pHash) was developed to survive benign transformations by fingerprinting visual or audio features rather than exact bytes. It is powerful for large-scale detection — think PhotoDNA or PDQ — but it carries its own evidentiary risks. Perceptual algorithms are susceptible to collision and targeted second-preimage attacks, in which an adversary manipulates invisible high-frequency noise to force a false match. The point was demonstrated publicly against Apple's NeuralHash model. As a result, courts are reluctant to accept a perceptual match alone as proof of copying; it is an investigative tool that needs cryptographic anchoring and human expert analysis behind it.

Possession is not authorship

The deepest limitation is doctrinal. A blockchain-anchored timestamp proves that a specific sequence of binary data existed on a specific date. It does not prove that the holder is the human author, nor that the data meets the threshold for copyrightability.

Under Feist Publications v. Rural Telephone Service (1991), protection extends only to original works of authorship — requiring at least a modicum of human creativity. Hash a public-domain dataset, a bare factual list, or a raw generative-AI output, and a timestamp proves only that you possessed the file. And under Computer Associates v. Altai (1992), two programmers may independently write nearly identical code when constrained by the same functional requirements. A timestamp can prove the plaintiff possessed the code first; it cannot prove access or intentional copying.

C2PA and the future of provenance

Recognising that a detached hash cannot bridge the gap between file integrity and human attribution, the industry is moving toward embedded provenance. The Coalition for Content Provenance and Authenticity (C2PA) defines an open specification that injects a cryptographically sealed manifest into a media file. A claim can carry assertions of identity and editing actions, a hard binding (a SHA-256 hash of the exact content) and a soft binding (a perceptual fingerprint or watermark that survives metadata stripping). Signing the manifest with a PKI certificate and anchoring it with an RFC 3161 timestamp produces a self-contained, tamper-evident record of who, when and what. That is precisely the combination that modern evidence rules reward — and it is why C2PA is emerging as the likely evidentiary gold standard.

The bottom line

Cryptographic hashing has genuinely solved the problem it set out to solve: digital anteriority and file integrity. What it has not solved is authorship, originality and jurisdiction. Cryptographic evidence proves that data existed at a moment in time; it cannot testify to who created it or whether the law protects it. The practical answer is layered: register where registration is required, anchor a cryptographic timestamp everywhere else, and understand the difference between having proof and having a claim.

Sources & further reading