Lösenordsstyrka-krav 2025: NIST-riktlinjer & bästa praxis

Password policies have undergone a radical transformation in recent years. The old rules — forced special characters, mandatory 90-day rotations, security questions — have been officially retired by NIST (National Institute of Standards and Technology). In their place, a new evidence-based approach focuses on length, breach-list checking, and multi-factor authentication. This comprehensive guide covers everything you need to know about modern password requirements in 2025, from the latest NIST SP 800-63B guidelines to practical implementation strategies.

1. NIST SP 800-63B: What Changed

NIST Special Publication 800-63B (Digital Identity Guidelines: Authentication and Lifecycle Management) was first revised in 2017 and updated through 2024. It fundamentally changed how organizations should think about password security. The key insight: most traditional password rules made passwords harder for humans but not meaningfully harder for attackers.

# NIST SP 800-63B Summary of Key Changes
┌─────────────────────────────────────────────────────────────────────┐
│  REMOVED (SHALL NOT)              │  ADDED (SHALL / SHOULD)         │
├───────────────────────────────────┼─────────────────────────────────┤
│  Forced periodic rotation         │  Check against breach lists     │
│  Composition rules (upper+lower   │  Minimum 8 chars (12+ better)   │
│    +digit+special)                │  Allow 64+ character passwords  │
│  Password hints                   │  Accept all Unicode + spaces    │
│  Security questions (KBA)         │  Allow paste in password fields │
│  Truncating passwords             │  Offer "show password" option   │
│  SMS as primary 2FA (restricted)  │  Use approved hash algorithms   │
└───────────────────────────────────┴─────────────────────────────────┘

2. Minimum Length: 8 Is Dead, 12-16 Is the New Standard

NIST SP 800-63B requires a minimum of 8 characters for user-chosen passwords, but security experts and modern frameworks have moved well beyond this floor. The mathematical reality is simple: every additional character exponentially increases the search space an attacker must cover.

Here is how password length affects brute-force time assuming 95 printable ASCII characters and an attacker hashing at 10 billion attempts per second (modern GPU cluster against a fast hash):

Modern recommendations by organization:

• NIST SP 800-63B: Minimum 8 characters (but recommends allowing up to 64+)
• OWASP: Minimum 8 characters, recommends 12+
• PCI-DSS v4.0: Minimum 12 characters (increased from 7 in v3.2)
• Microsoft: Minimum 12 characters for Azure AD
• Google: Minimum 8 characters, recommends 12+

3. What NIST Recommends: Allow All Characters, Check Breach Lists, No Hints

The NIST guidelines take a fundamentally different approach than traditional password policies. Instead of making passwords hard to create, they focus on making passwords hard to guess.

4. What NIST Removed: Forced Special Characters, Periodic Expiration, Security Questions

Understanding what was removed is just as important as understanding what was added. Each removed requirement was based on assumptions that research proved wrong.

5. Entropy Calculator: How to Estimate Password Strength in Bits

Password entropy measures the unpredictability of a password in bits. Higher entropy means the password is harder to guess. Entropy is calculated as: H = log2(R^L), where R is the size of the character pool and L is the password length.

# Password Entropy Formula
H = log2(R^L) = L * log2(R)

Where:
  H = Entropy in bits
  R = Size of the character pool
  L = Password length

# Example: 12-character password using full ASCII (95 chars)
H = 12 * log2(95) = 12 * 6.57 = 78.8 bits

# Example: 4-word Diceware passphrase (7776-word list)
H = 4 * log2(7776) = 4 * 12.92 = 51.7 bits

Character Pool Sizes

Entropy Calculations for Common Passwords

Target entropy levels for 2025:

• Online attacks (rate-limited): 30+ bits is sufficient with account lockout
• Offline attacks (stolen hash, fast hash): 60+ bits minimum
• Offline attacks (stolen hash, Argon2id): 40+ bits is acceptable
• High-security accounts: 80+ bits recommended

6. Comparison Table: NIST vs OWASP vs PCI-DSS vs Old Standards

Different standards and frameworks have different requirements. Here is how they compare as of 2025:

7. Passphrase vs Password: Why "correct horse battery staple" Wins

The famous XKCD comic (#936) made a compelling case for passphrases over passwords. The math supports it: a randomly selected 4-word passphrase from a 7,776-word Diceware list has 51.7 bits of entropy — comparable to a random 8-character mixed-case alphanumeric password — while being dramatically easier to remember.

Passphrase vs Traditional Password Comparison

Why passphrases are better:

• Equal or greater entropy: A 4-word passphrase matches an 8-character random password in entropy. A 6-word passphrase exceeds most traditional passwords.
• Easier to remember: Humans are better at remembering sequences of words than random characters. A passphrase creates a mental image.
• Easier to type: Especially on mobile devices, typing words is faster and less error-prone than switching between character types.
• Resistant to shoulder surfing: A long string of words is harder to memorize from a brief glance than a short random string.
• Naturally long: Even a 4-word passphrase is typically 20+ characters, providing excellent protection against brute-force attacks.

How to generate a Diceware passphrase:

1. Get a Diceware word list (7,776 words, each mapped to a 5-digit dice roll).
2. Roll 5 dice (or one die 5 times) for each word you need.
3. Look up each 5-digit number in the word list.
4. Combine the words with spaces. That is your passphrase.
5. For most purposes, 5-6 words provide excellent security (64.6 - 77.5 bits).

# Diceware passphrase generation example
$ dice rolls: 1-6-2-3-4  →  "atlas"
$ dice rolls: 5-4-1-1-6  →  "usable"
$ dice rolls: 2-5-3-1-2  →  "flock"
$ dice rolls: 2-1-4-6-5  →  "dizzy"

Passphrase: "atlas usable flock dizzy"
Entropy: 4 * log2(7776) = 51.7 bits
Length: 24 characters (spaces included)

# Compare to traditional password
"j7#Kp!2x" = 8 chars * log2(95) = 52.6 bits
# Same entropy, but much harder to remember and type!

8. Implementing Password Policies: Bcrypt Cost, Rate Limiting, MFA

Knowing the guidelines is one thing; implementing them correctly is another. Here is a practical checklist for implementing a modern, NIST-compliant password policy in 2025.

Password Hashing

Choose Argon2id as your primary hashing algorithm. If Argon2id is not available, use bcrypt with a cost factor of 12 or higher. Target 200-500ms per hash on your production hardware. Never use MD5, SHA-1, SHA-256, or any fast hash function for password storage.

// Argon2id configuration (OWASP recommended)
const argon2 = require('argon2');

const hashConfig = {
  type: argon2.argon2id,
  memoryCost: 19456,    // 19 MiB
  timeCost: 2,          // 2 iterations
  parallelism: 1,       // 1 thread
};

// Hash on registration
const hash = await argon2.hash(password, hashConfig);

// Verify on login
const isValid = await argon2.verify(hash, password);

// ── Bcrypt alternative (if Argon2 unavailable) ──
const bcrypt = require('bcrypt');
const COST_FACTOR = 12;  // Target: 200-500ms

const bcryptHash = await bcrypt.hash(password, COST_FACTOR);
const bcryptValid = await bcrypt.compare(password, bcryptHash);

Rate Limiting

Implement rate limiting on authentication endpoints to prevent online brute-force attacks. After 5-10 failed attempts, introduce progressive delays or temporary account lockout. Use a CAPTCHA after 3 failed attempts. Implement IP-based rate limiting (e.g., 100 attempts per IP per hour) and account-based rate limiting (e.g., 10 attempts per account per 15 minutes).

// Rate limiting example (Express.js with express-rate-limit)
const rateLimit = require('express-rate-limit');

// IP-based rate limiting
const loginLimiter = rateLimit({
  windowMs: 15 * 60 * 1000,  // 15 minutes
  max: 10,                     // 10 attempts per IP per window
  message: 'Too many login attempts. Please try again later.',
  standardHeaders: true,
  legacyHeaders: false,
});

// Account-based rate limiting (in-memory example)
const accountAttempts = new Map();

async function checkAccountRateLimit(username) {
  const key = username.toLowerCase();
  const attempts = accountAttempts.get(key) || { count: 0, lastAttempt: 0 };

  if (attempts.count >= 5) {
    const cooldown = 15 * 60 * 1000; // 15 minutes
    if (Date.now() - attempts.lastAttempt < cooldown) {
      throw new Error('Account temporarily locked. Try again later.');
    }
    attempts.count = 0; // Reset after cooldown
  }

  return attempts;
}

app.post('/login', loginLimiter, async (req, res) => {
  const { username, password } = req.body;
  const attempts = await checkAccountRateLimit(username);

  const isValid = await verifyPassword(username, password);

  if (!isValid) {
    attempts.count++;
    attempts.lastAttempt = Date.now();
    accountAttempts.set(username.toLowerCase(), attempts);
    return res.status(401).json({ error: 'Invalid credentials' });
  }

  accountAttempts.delete(username.toLowerCase());
  // ... generate session / JWT
});

Multi-Factor Authentication (MFA)

MFA is the single most effective defense against credential-based attacks. Even if an attacker obtains a password (through phishing, breach reuse, or brute force), MFA blocks unauthorized access. Prefer app-based TOTP (Google Authenticator, Authy) or hardware keys (YubiKey, Passkeys) over SMS-based 2FA.

Server-Side Validation Checklist

Bcrypt / Argon2id Cost Tuning

The cost parameter should be tuned to your specific hardware. Run benchmarks on your production server and choose a cost that results in 200-500ms per hash. This provides a good balance between user experience (login latency) and attacker cost.

# Benchmarking password hashing on your server
# Goal: 200-500ms per hash

# Node.js benchmark script
const argon2 = require('argon2');
const bcrypt = require('bcrypt');

async function benchmark() {
  const password = 'benchmarkPassword123';

  // Argon2id benchmarks
  for (const mem of [19456, 47104, 65536]) {
    const start = Date.now();
    for (let i = 0; i < 10; i++) {
      await argon2.hash(password, {
        type: argon2.argon2id,
        memoryCost: mem,
        timeCost: 2,
        parallelism: 1,
      });
    }
    const avg = (Date.now() - start) / 10;
    console.log(`Argon2id (m=${mem}): ${avg.toFixed(0)}ms`);
  }

  // Bcrypt benchmarks
  for (const cost of [10, 11, 12, 13, 14]) {
    const start = Date.now();
    for (let i = 0; i < 10; i++) {
      await bcrypt.hash(password, cost);
    }
    const avg = (Date.now() - start) / 10;
    console.log(`Bcrypt (cost=${cost}): ${avg.toFixed(0)}ms`);
  }
}

benchmark();

# Expected output (varies by hardware):
# Argon2id (m=19456): ~150ms
# Argon2id (m=47104): ~350ms   ← Good target
# Argon2id (m=65536): ~500ms
# Bcrypt (cost=10):   ~75ms
# Bcrypt (cost=11):   ~150ms
# Bcrypt (cost=12):   ~300ms   ← Good target
# Bcrypt (cost=13):   ~600ms
# Bcrypt (cost=14):   ~1200ms

9. Checking Against Breach Databases: Have I Been Pwned API, k-Anonymity

One of NIST's most impactful requirements is checking passwords against known breach databases. The Have I Been Pwned (HIBP) Passwords API makes this both practical and privacy-preserving using a technique called k-anonymity.

How k-Anonymity Works

The HIBP Passwords API never sees the actual password or even the full hash. Instead, it uses a range search based on the first 5 characters of the SHA-1 hash:

1. Hash the password with SHA-1 (used only for lookup, not for storage).
2. Send only the first 5 characters of the hex-encoded hash to the HIBP API.
3. The API returns all hash suffixes that match the 5-character prefix (typically 500-800 results).
4. Compare the full hash against the returned suffixes locally.
5. If a match is found, the password has been seen in a breach. Reject it.

// Check password against Have I Been Pwned API (Node.js)
const crypto = require('crypto');

async function isPasswordBreached(password) {
  // Step 1: SHA-1 hash the password
  const sha1 = crypto.createHash('sha1')
    .update(password)
    .digest('hex')
    .toUpperCase();

  // Step 2: Split into prefix (5 chars) and suffix
  const prefix = sha1.substring(0, 5);
  const suffix = sha1.substring(5);

  // Step 3: Query the HIBP API with only the prefix
  const response = await fetch(
    `https://api.pwnedpasswords.com/range/${prefix}`,
    {
      headers: { 'Add-Padding': 'true' }  // Prevents response-length analysis
    }
  );
  const text = await response.text();

  // Step 4: Check if our suffix appears in the results
  const lines = text.split('\n');
  for (const line of lines) {
    const [hashSuffix, count] = line.split(':');
    if (hashSuffix.trim() === suffix) {
      return {
        breached: true,
        count: parseInt(count.trim(), 10),  // How many times seen
      };
    }
  }

  return { breached: false, count: 0 };
}

// Usage in password validation
async function validatePassword(password, username) {
  // Check length
  if (password.length < 12) {
    return { valid: false, reason: 'Password must be at least 12 characters.' };
  }

  // Check against username / email
  if (password.toLowerCase().includes(username.toLowerCase())) {
    return { valid: false, reason: 'Password must not contain your username.' };
  }

  // Check against breach database
  const breach = await isPasswordBreached(password);
  if (breach.breached) {
    return {
      valid: false,
      reason: `This password has appeared in ${breach.count.toLocaleString()} data breaches. Please choose a different password.`,
    };
  }

  return { valid: true };
}

As of 2025, the HIBP Passwords database contains over 900 million unique compromised passwords. If a user tries to set a password that appears in this list, it should be rejected with a clear explanation.

Alternatives to HIBP

10. Password Manager Recommendations

With passwords becoming longer and more unique (never reused across sites), password managers are no longer optional — they are essential. A password manager generates, stores, and auto-fills strong unique passwords for every account.

Why password managers are critical:

• The average person has 100+ online accounts. No one can remember 100 unique, strong passwords.
• Password reuse is the #1 cause of credential-based breaches. A password manager eliminates reuse entirely.
• Password managers generate truly random passwords with maximum entropy, not human-predictable patterns.
• Auto-fill prevents phishing: a password manager will not fill credentials on a lookalike phishing domain.

11. Frequently Asked Questions