AI-enabled social engineering

AI has not invented a new attack. It has made the old one cheap, fast and well targeted.

Phishing used to carry tells, because writing a thousand convincing, personalised messages took effort nobody had. That constraint is gone. The spelling is right, the tone matches your company, the details are drawn from your own public footprint, and it costs almost nothing to produce at scale. Advice built on spotting bad English no longer does much. What still holds is that verification through a separate channel does not care how convincing the message was.

Checked against the primary source.

More on AI security