Leaving this course?

Lesson 1 of 5

How passwords are actually broken

Almost nobody has a password guessed by somebody typing at a login screen. The real methods are different, and knowing them changes what good advice looks like.

Reuse, and the breach of somewhere else. Billions of username and password pairs from past breaches are available to attackers. They try them in bulk against other services, which works because most people reuse. Your password's strength is irrelevant here: it was not broken, it was taken from somewhere that lost it.

Guessing at scale, offline. When a service is breached, attackers take the stored password data and try enormous numbers of candidates against it on their own hardware, with no rate limit and no lockout. Common words, names, dates and predictable substitutions fall quickly.

Phishing. The user types the password into a convincing copy of the real site. No guessing, no breach: it is simply handed over. This is the method that defeats password strength entirely.

Spraying. Rather than many guesses against one account, one or two very common passwords against thousands of accounts. It avoids lockouts, and in an organisation of any size somebody is using the season and the year.

What this means for advice. Length beats complexity, because length is what makes offline guessing expensive. Uniqueness beats both, because it is the only defence against a breach elsewhere. And neither helps against phishing, which is what the second and third lessons are about.

The advice that has changed. Forced regular changes are now discouraged by major guidance, because they push people toward predictable patterns — the same word with an incrementing number. Change a password when there is reason to think it is compromised, not on a calendar.

Lesson 2 of 5

What makes a good password, and the manager question

Three properties matter, in this order: it is unique to one service, it is long, and it is not a predictable word or pattern.

Unique first. One password per service is what stops a breach somewhere else becoming a problem everywhere. Nothing else achieves this, and no amount of strength substitutes for it.

Then long. Three or four unrelated words are both memorable and expensive to guess — far more so than a short string of substituted characters. Complexity requirements that force a symbol and a digit mostly produce the same password with a 1 and an exclamation mark on the end.

Then unpredictable. Not the organisation's name, not the month, not a child's name and a birth year, not a keyboard pattern. Anything derivable from the person or the place will be tried first.

Which leads to the obvious problem. Nobody can remember thirty unique long passwords, and that is not a failure of discipline — it is the reason reuse is universal.

So use a password manager. It generates unique passwords, stores them, and fills them in. The common objection is that it puts everything in one place, and the honest answer is that the alternative in practice is reuse, which is worse. A manager also resists phishing in one useful way: it will not fill a password into a site whose address does not match.

Protect the manager properly. One long unique passphrase that is not used anywhere else, plus a second factor on the manager itself. That is the one password worth memorising.

And write the few that matter on paper if you need to. A recovery passphrase kept in a locked drawer is a reasonable control against the risk of losing everything. The attacker in your threat model is not usually somebody who can open your desk.

Lesson 3 of 5

The second factor, and which kinds are weak

A second factor means a stolen password is not enough on its own. That is a large improvement, and the kinds differ enormously in what they actually resist.

Codes by text message. Better than nothing, and the weakest kind. The number can be moved to an attacker's control by persuading the mobile operator — the attack usually called a SIM swap — and the code can be phished alongside the password, because the user will type it into the fake site as readily as the password.

Codes from an app. A six-digit code that changes every thirty seconds, generated on the device. Immune to SIM swapping, still phishable: a convincing fake site asks for the code and relays it within the window.

Push approvals. A prompt on the phone saying "approve this sign-in". Convenient, and vulnerable to fatigue: an attacker with the password sends prompts repeatedly until somebody taps approve to make it stop. Number matching, where the site shows a number that must be typed into the phone, substantially fixes this.

Hardware keys and passkeys. The strong kind. The device proves possession cryptographically and — the important part — checks the address of the site it is talking to. A fake site cannot obtain anything reusable, which is why this category is described as phishing-resistant and the others are not.

The ranking, plainly. Any second factor beats none. App codes beat text messages. Push with number matching beats plain push. Hardware keys and passkeys beat everything, and are the only kind that stops phishing rather than making it harder.

And put it where it matters most. Email first, because email resets everything else. Then anything with money or administrative control. An account whose recovery is an unprotected mailbox is only as strong as that mailbox.

Lesson 4 of 5

How passkeys work

Passkeys are the first change in a long time that removes the password rather than defending it, and the idea underneath is simpler than the marketing suggests.

Two keys instead of one secret. When you register, your device creates a pair: a private key that never leaves it, and a public key that goes to the service. To sign in, the service sends a challenge, the device signs it with the private key, and the service checks the signature with the public key.

The service never holds anything worth stealing. A public key is useless to an attacker. A breach of the service no longer leaks credentials that work anywhere, which removes the entire reuse problem at a stroke.

And there is nothing to phish. The device will only sign for the site it registered with, checked by its actual address. A convincing copy at a lookalike address gets nothing, because the browser and the device will not produce a signature for it. This is the property that matters most.

Unlocking is local. The fingerprint, face or device PIN unlocks the private key on the device. That biometric does not go to the service and is not stored by it.

Where they live. Either synchronised across your devices through a platform or password manager account, or bound to one physical device such as a security key. Synchronised is more convenient and makes the account protecting the sync important; device-bound is stronger and loses access if the device is lost.

Recovery is the hard part. If the sign-in method is on a device, losing the device matters. Register more than one — a second device, or a hardware key kept somewhere safe — before you need to, because the recovery path is the weakest link in any authentication system and attackers know it.

They are not universal yet. Many services still want a password as well, so for now most people will hold both, and the password still needs to be unique and long.

Lesson 5 of 5

What an organisation should do

The individual advice is simple. Making it true across an organisation is a different problem, and a few decisions carry most of the benefit.

Turn on multi-factor authentication everywhere reachable from outside. Email, remote access, administrative consoles, cloud management. This removes the most common intrusion route, and it is the single highest-value control in this course.

Prefer phishing-resistant factors for the accounts that matter. Administrators, finance approvers, anybody who can move money or change access. Hardware keys or passkeys rather than codes, because those roles are the ones worth a targeted attempt.

Stop forcing routine password changes, and stop requiring complexity patterns that produce predictable results. Set a long minimum length instead, and block the passwords that appear in breach lists. Several services will do that check for you.

Give people a password manager, as a provided tool rather than something they are left to arrange. Reuse is a consequence of an impossible requirement, and the tool removes the requirement.

Fix the recovery path. Help desk resets are a common way in: an attacker calls, sounds plausible, and is given access. Decide how identity is confirmed before a reset, and make it the same for everybody including senior people, because the pressure to make exceptions is where this fails.

Watch the exceptions you have already granted. Shared accounts, service accounts with passwords in scripts, the system that could not support a second factor two years ago. These accumulate, and they are where an attacker ends up.

And measure the coverage. What proportion of accounts have a second factor, which systems still accept a password alone, and how many accounts are exempt. Three numbers most organisations cannot produce, and the honest starting point.

Knowledge check

Ten questions

Answer all ten questions, then check your answers. You need 9 out of 10 to pass and receive a certificate. If you score less, you will see which answers were right and wrong, and then go through the course again before you retake the check. Your answers, progress and times are kept only in this browser.

Standards and frameworks covered

The published frameworks this course teaches toward. Saolix is not affiliated with, endorsed by or accredited by any of these bodies; naming a standard is a citation, not a claim about its author.

Sources

The official documents this course relies on. Laws and guidance change, so check the current version.

  1. Password administration for systems owners · National Cyber Security Centre, United Kingdom
  2. Digital Identity Guidelines (SP 800-63) · National Institute of Standards and Technology
  3. Multi-factor authentication guidance · Cybersecurity and Infrastructure Security Agency, United States
  4. Web Authentication (WebAuthn) specification · World Wide Web Consortium
  5. Passkeys · FIDO Alliance
  6. Framework for Improving Critical Infrastructure Cybersecurity · National Institute of Standards and Technology
  7. Computer Misuse and Cybercrimes Act, 2018 (No. 5 of 2018) · Kenya Law
  8. Data Protection Act, 2019 (No. 24 of 2019) · Kenya Law
  9. ISO/IEC 27001 Information security management systems · International Organization for Standardization
  10. OWASP Authentication Cheat Sheet · OWASP Foundation