The biggest change AI brings to phishing may not be a brand-new type of scam.
It may simply make old scams easier to run.
A criminal can use AI to produce convincing messages, translate them, personalise them and generate multiple variations quickly.
That means phishing does not need to look obviously amateur anymore.
Users should therefore focus on the request rather than the writing quality.
Unexpected account warnings, refund notices, delivery problems, payment requests and password-reset instructions deserve independent verification.
If the message asks you to sign in, do not follow the supplied link. Open the service yourself.
If the message asks for money, verify the request using another communication channel.
AI can improve the scam's presentation. It cannot make an independently verified request fraudulent-proof.