AI PASSWORDS
Chatbots are not your password manager
If you’ve ever asked AI to generate a password, this is worth a quick pause.
Cybersecurity researchers are warning that chatbots like ChatGPT, Claude, and Gemini aren’t reliable tools for creating secure passwords.
The reason is simple: these models don’t produce true randomness.
Instead, they generate passwords that look strong, but often follow predictable patterns.
In some tests, different accounts were shown producing the same or nearly identical passwords, making them easier to crack than they appear.
Here’s what you should know:
AI passwords can look strong but still be predictable
Password managers remain the safest choice
Weak credentials may slip into AI-generated code without warning
Strong-Looking ≠ Strong
Security expert Dan Laav, who works with major AI labs and governments, says this is a growing issue because people may not even realise they’re exposed.
Weak AI-generated passwords can also end up inside AI-written code, meaning insecure credentials could be published online without anyone noticing.
Even the models themselves admit the limitation: they’re generating text based on training patterns, not using cryptographic tools designed for real password security.
The safer option is still the boring one: password managers and passkeys, which rely on proper random number generation.
So if you’ve used AI to make a password before, experts recommend changing it.
I just headbutt the keyboard to generate mine. - MV