AI-Written Phishing

Phishing used to be reliably spottable by bad grammar, generic greetings, and obviously wrong sender details. AI language tools have removed most of that: scammers can now generate flawless, personalized messages that closely mimic a real company's tone, reference specific details scraped from public sources, and arrive with correct-looking sender names.

This means the old advice — "look for typos" — is no longer sufficient advice on its own. The messages that matter most now look exactly like legitimate correspondence, sometimes referencing a real recent order, a real coworker's name, or a real upcoming deadline to feel contextually accurate.

What still works: checking the actual sender address (not just the display name), never clicking a link in an unexpected message when you can navigate to the site directly instead, and treating any message that creates urgency around credentials or payment with the same suspicion regardless of how polished it looks.

• The giveaway: No longer bad grammar — now it's urgency plus a request for credentials, payment, or a click.
• What to do: Check the real sender address, and navigate to sites directly instead of clicking links.
• Red flag: Any message creating pressure to act on credentials or money right now.
• When in doubt: Contact the company or person through a channel you already know is real.