Hash Value and the Integrity of Electronic Evidence under the Bharatiya Sakshya Adhiniyam, 2023

Paper shows its own history. Erasures disturb the fibres, interpolations sit in different ink, and an added page is often a different sheet. Electronic records show nothing at all: a file altered a moment ago looks exactly like one untouched for years. The hash value is the answer the discipline has developed to this problem, and the Adhiniyam has made its disclosure a statutory requirement in Part A of the Schedule. It is the closest thing electronic evidence has to the physical integrity of a document.

1. The Statutory Requirement

The Schedule requires Part A of the certificate to disclose the hash value of the electronic record, obtained through one of the algorithms it specifies. The algorithms named are the standard families in use — SHA-1, SHA-256 and MD5.

This requirement is entirely new. Section 65B(4) of the Indian Evidence Act prescribed no form for the certificate and said nothing about hash values. Its introduction was one of the two grounds on which the Schedule was challenged, and the challenge failed.

The Court's characterisation of electronic records as 'liable to continuous mutation' is the key to why the requirement exists. A paper document is stable and its alteration leaves traces. An electronic record is not stable, and its alteration leaves none. Something had to take the place of the physical integrity a paper original supplies, and the hash value is that something.

2. What a Hash Value Is

A cryptographic hash function takes a file of any size and produces a fixed-length output — the hash value or digest — usually expressed as a string of hexadecimal characters. Four properties give it its evidentiary use.

Determinism. The same input always produces the same output. A file hashed today and hashed again in five years yields the same value, so a record can be verified at any point.

The avalanche effect. Any change to the input, however small, produces a completely different output. Altering a single character, a single pixel, or a single bit does not produce a slightly different hash; it produces one bearing no resemblance to the original. There is no such thing as a nearly matching hash.

Fixed length. The output is the same length whatever the input. A one-page document and a ten-gigabyte disk image both produce a digest of the same size, which makes hashes practical to record in a certificate or a seizure memo.

Practical irreversibility. The function cannot be run backwards to reconstruct the file from the digest, and it is computationally impractical to construct a different file producing the same digest.

3. The Three Algorithms

The Schedule names three algorithms, and they are not of equal strength. The relevant property is collision resistance — the difficulty of constructing two different files that produce the same digest.

The distinction matters in a specific and limited way. All three algorithms detect alteration of an existing file with equal reliability: if a record is changed after hashing, the digest changes, whichever algorithm was used. The weakness of MD5 and SHA-1 lies in collision construction — the possibility of deliberately creating two different files sharing a digest, which is a different attack and requires the adversary to control both files from the outset.

For most litigation this is theoretical, and an MD5 hash recorded at seizure remains good evidence that the record has not been altered since. Where fabrication by a technically capable party is genuinely in issue, the point is available and the stronger algorithm should be used. The practical advice is simply to use SHA-256, or to record more than one digest, which costs nothing.

4. When the Hash Must Be Computed

This is the single most important practical point in the subject, and it is where cases are lost.

A hash computed at the time of tendering proves that the record has not changed since the party decided to produce it. That establishes almost nothing, because the period that matters is the one between the event and the tender.

A hash computed at the time of seizure or acquisition proves that the record produced in court is identical to the record as it existed when it came into the hands of the person producing it. That is the assurance the requirement exists to provide.

The points at which a hash should be computed are therefore the points at which custody changes or a copy is made.

1. At seizure, when a device or storage medium is taken into possession. The digest should be recorded in the seizure memo or panchnama and attested by the witnesses to the seizure.
2. At forensic imaging, when a bit-for-bit copy of the storage is made. The digest of the source and of the image should both be computed and shown to match.
3. At download or extraction, where a record is obtained from a service provider, a portal or an application, so that the state of the record at the moment of acquisition is fixed.
4. At each transfer, where the record passes from one custodian to another, so that the point of any discrepancy can be identified.
5. At examination, so that the examiner can show he worked on the material he was given and returned it unchanged.
6. At production, where the output tendered is generated, so that the certificate can identify what is being produced.

5. Hashing in Forensic Practice

The hash does not operate alone. It is one element of a set of practices developed to preserve electronic material, and the others matter as much.

Write blocking. A storage device connected to a computer is written to in the ordinary course — timestamps update, indexes are created, temporary files appear. A write blocker is hardware or software preventing any write operation to the source, so that the act of examining a device does not alter it. Without a write blocker the digest of the source will change simply by being examined, and the party will be unable to explain the discrepancy.

Forensic imaging. A bit-for-bit copy of the entire storage, including unallocated space and slack space where deleted material persists. The image is hashed, the source is hashed, and the two are shown to match. All subsequent examination is conducted on a working copy of the image, leaving both the source and the master image untouched.

Verification at each stage. The digest is recomputed at each transfer and each stage of examination, so that if a discrepancy arises the point at which it arose can be identified.

Documentation. The digests, the tools used and their versions, the times, and the persons involved are recorded contemporaneously, so that the chain can be described in evidence.

The Bharatiya Nagarik Suraksha Sanhita, 2023 now supports these practices in two ways relevant here. Section 105 requires the audio-video recording of search and seizure, which produces a visual record of what was seized and in what condition. Section 176(3) requires, for offences punishable with seven years or more, that a forensic expert visit the scene and that the collection be videographed. Where these are complied with, the circumstances of acquisition become verifiable in a way that supports the digest rather than depending on it.

6. The Chain of Custody Gap

The hash requirement addresses, in part, a gap the Adhiniyam otherwise leaves open.

The Standing Committee on Home Affairs, examining the Bharatiya Sakshya Bill, observed that it contained no provision securing the chain of custody of material and digital records collected during investigation. That observation remains accurate. The Adhiniyam says how a record becomes admissible; it says nothing about how its integrity between seizure and production is to be maintained or established.

The hash disclosure in Part A of the Schedule is the nearest the statute comes to filling the gap, and it fills it only partially. It establishes that the record produced corresponds to the record as hashed. It says nothing about who held the record in the interval, whether it was accessible to persons with an interest, or whether the digest itself was recorded reliably.

The practical consequence is that a party relying on electronic evidence must build the chain of custody by ordinary evidence — the seizure memo, the register, the transfer receipts, the testimony of custodians — and use the hash as the technical anchor within it, rather than expecting the hash to do the whole work.

7. Verification and Challenge in Court

A hash disclosed in a certificate is capable of being tested, and this is one of its considerable advantages over most technical assertions.

For the party producing the record, the strongest position is a digest recorded in the seizure memo, attested by independent witnesses, matching the digest in Part A of the certificate, matching in turn the digest of the file actually filed in court. Where the three correspond, the objection that the record has been altered is answered conclusively, and no amount of cross-examination will disturb it.

For the party challenging, several lines are available and they should be taken in order.

• When was the digest first computed? If only at the time of certification, the intervening period is unaccounted for.
• Who computed it, and how is it recorded? A digest appearing for the first time in the certificate, with no contemporaneous record, is an assertion rather than a verification.
• What is the digest of? The source device, the forensic image, the extracted file, or the output tendered — these are different values, and a certificate that does not say which is unclear.
• Was a write blocker used? If not, the source may have been altered by the examination itself.
• Does the digest match on verification? The court may direct recomputation, and the exercise is quick and conclusive.
• Which algorithm was used? Where fabrication rather than alteration is alleged, the weaker algorithms are open to the collision argument.

8. What a Hash Does Not Establish

The digest answers one question with near-certainty and answers no others at all. Four limits should be kept firmly in view, because a matching hash is easily over-read.

It does not establish authenticity. A hash proves that a file has not changed since it was hashed. It says nothing about whether the file was genuine when it was hashed. A fabricated record, hashed at the moment of fabrication and produced with a matching digest, will verify perfectly. The digest fixes a state; it does not validate it.

It does not establish attribution. That a message file has not been altered says nothing about who sent the message. Attribution is to a device or an account, and who was operating it must be established by other evidence. Section 90 expressly declines to presume the sender of an electronic message.

It does not establish truth. Section 63(1) admits computer output as evidence of a fact only 'of which direct evidence would be admissible'. A statement that would be hearsay from a witness remains hearsay in a verified file.

It does not establish completeness or meaning. A hash of three messages extracted from a thread of three hundred verifies the three. Whether enough has been produced for the meaning to be understood is governed by Section 33, and what the fragment signifies is a matter for expert opinion under Section 39(2).

9. The Position Stated Shortly

1. The hash requirement is new, introduced by the Schedule, and was upheld in Pune Bar Assn. as bearing a rational nexus with securing authenticity and integrity.
2. A hash is deterministic and shows the avalanche effect — any change, however small, produces a completely different digest. There is no partial match.
3. The Schedule names SHA-1, SHA-256 and MD5, and SHA-256 is the algorithm to use where a choice exists.
4. All three detect alteration equally. The weakness of MD5 and SHA-1 lies in collision construction, which is a different and rarer attack.
5. The digest must be computed at seizure or acquisition, not at tender, and recorded in the seizure memo so its provenance is independent of the certificate.
6. A hash never taken cannot be supplied later, and the objection that a record may have been altered then cannot be answered at all.
7. The hash works with write blocking, forensic imaging and documentation, and does not operate alone.
8. A hash establishes integrity only. It does not establish authenticity, attribution, truth, completeness or meaning.

10. Related Topics and Provisions