Hash Generator

For integrity checks and general-purpose hashing, SHA-256 is the sensible default. MD5 and SHA-1 are both cryptographically broken — practical collision attacks exist for each, and SHA-1 collisions were demonstrated publicly in 2017 — so neither should back a security decision, though MD5 remains acceptable as a fast non-security checksum for cache keys or deduplication. For hashing passwords, none of these are appropriate.

Because SHA-256 is designed to be fast, and speed is exactly what an attacker with a stolen database wants — commodity GPUs compute billions of SHA-256 hashes per second, making brute-force and rainbow-table attacks cheap. Password storage needs a deliberately slow, salted, memory-hard function: bcrypt, scrypt, or Argon2. These accept a work factor you raise as hardware improves, and they salt automatically so identical passwords produce different stored hashes.

A collision occurs when two different inputs produce the same digest. Collisions must exist mathematically, since a fixed-length output cannot uniquely represent unbounded input — what matters is whether they can be found deliberately. For MD5 and SHA-1 they can, cheaply, which lets an attacker craft two files with matching digests and substitute one for the other. SHA-256 has no known practical collision attack.

Almost always an input difference rather than an algorithm one. Trailing newlines, a UTF-8 byte-order mark, and CRLF versus LF line endings all change the bytes and therefore the digest. Text hashed as a string also differs from the same text hashed as an uploaded file if encoding or line endings differ. Comparing bytes rather than assuming visual equivalence resolves most mismatches.

Not by inverting the function — hashes are one-way by construction. But they are not secret: identical input always yields identical output, so an attacker can hash likely candidates and compare. That is precisely how rainbow tables and dictionary attacks recover common passwords from unsalted digests. A hash protects data only when the input has enough entropy that guessing it is infeasible.