Password work in a forensic lab is engineered, budgeted and reported like any other stage. The examiner extracts the target hash or container, characterises the encryption, and applies a documented sequence of attacks against it on dedicated GPU hardware. Nothing is guessed at; nothing is off-the-shelf-magic. The report records what was tried, in what order, on what hardware, and for how long — because a negative result is a result, and it has to be defensible.
How the work is actually done
- 01Characterise the target. Identify the format (BitLocker, FileVault2, VeraCrypt, iOS backup, Office 2019+, 7-Zip, KeePass, etc.), the key derivation function, and the resulting hash type. This decides the attack tempo before any hardware is committed.
- 02Extract the hash or container. Isolate the material that actually needs to be attacked — a hash string, an encrypted header, or the container itself — without altering the source.
- 03Build an intelligent wordlist. Merge public breach corpora with client-supplied context: known passwords, family and pet names, dates, employers, project codenames, prior credentials from the same custodian.
- 04Apply rules and masks. Rule-based mutations of the wordlist (capitalisation, digit-append, leetspeak), then mask attacks constrained to the custodian's known patterns.
- 05Brute-force where feasible. GPU-accelerated exhaustion of the remaining keyspace on Hashcat or John the Ripper, on multi-GPU rigs sized to the format's tempo.
- 06Report. Attempts, tempo, coverage, elapsed time, and either the recovered credential or a defensible negative — with an estimate of what further work would cost and yield.
What actually yields — and what does not
| Target | Realistic outcome | Note |
|---|---|---|
| Older Office (≤2003) documents | Recovered in minutes to hours. | 40-bit encryption; effectively broken by design. |
| Legacy ZIP / RAR archives | Often recovered. | Weak KDF; dictionary + rules usually sufficient. |
| Office 2019 / 2021 documents | Depends on password strength. | Slow KDF; strong passwords may be out of reach. |
| 7-Zip / modern VeraCrypt containers | Slow — only feasible with intelligent wordlists. | Deliberately slow KDFs; brute force alone is impractical. |
| iOS backup (encrypted) | Depends on iOS version and password entropy. | Modern iOS backups use very slow KDF; wordlist quality is decisive. |
| BitLocker / FileVault2 without recovery key | Rarely recoverable against a strong passphrase. | TPM-backed; hardware protections make brute force uneconomic. |
| Modern Android with hardware-backed keystore | Not recoverable at the credential level. | Attack the extraction pathway, not the password. |
The five-minute conversation at intake
The single largest determinant of success is the wordlist. A short conversation with the custodian — pets, children, dates of birth, prior employers, project names, the pattern they know they use — produces a candidate list that a generic public wordlist will not touch. In practice, half of recoverable credentials fall to a wordlist built from the custodian's own history within the first hour of GPU time.
The right question is rarely 'can you break this?' — it is 'given the format, the hardware, and what the custodian will tell us about their own patterns, what is the probability of success, at what cost, in what time?' That is what the report should answer at the outset.
Legality and instruction
- The examiner must have written authority from the data controller or the court to attempt recovery.
- Attempts against systems the client does not own — third-party accounts, ex-spouse devices without an order — are not undertaken.
- Where the credential is recovered, it is disclosed only to the instructing party and recorded in the case file; it is never re-used or retained beyond the matter.
- Negative results are reported with the same rigour as positive ones, including the attack tempo achieved and the residual keyspace.
Frequently asked questions
Not by attacking the credential in isolation — the hardware-backed KDF makes brute force uneconomic. Access is normally achieved through the extraction pathway (see the mobile extraction guides), where the passcode is applied against the device itself under supervised conditions.
If the organisation is the data controller, the recovery key is almost always retrievable from Active Directory or Entra ID — no cracking required. Attacking the passphrase directly against a TPM-backed volume is rarely economic.
For weak formats, minutes to hours. For modern slow-KDF formats with intelligent wordlists, up to a few days on multi-GPU hardware. Beyond that, additional runtime buys diminishing returns and the report will say so.
Yes, when instructed by the data controller or under a court order, and on material the instructing party is entitled to access. The examiner will decline instructions that do not meet those tests.
The target file or image, a description of the custodian and context, and — most valuable — a brief conversation with someone who knows the custodian's password patterns. That single input decides most cases.