1. The Scam Economics Calculator
Sending email is nearly free, so even absurdly low response rates print money. Drag the sliders and watch the profit line. Making the email more polished raises reply volume, but it also raises the cost of handling replies from people who will never pay.
2. The Self-Selection Funnel
A famous Microsoft Research finding: scammers who say "Nigerian prince" up front are deliberately repelling skeptics. Everyone who replies has already ignored a giant red flag, so replies convert far better. Toggle the email style and watch the funnel reshape.
3. Five Reasons the Junk Never Stops
Advanced AI exists on both sides of the fight, and the sloppy stuff survives anyway. Here is why.
The typos are the filter
Obvious scams pre-qualify victims. Anyone who replies past bad grammar is unusually credulous, which is exactly who a scammer can afford to spend time on.
Marginal cost is near zero
Ten million emails cost a few dollars. A 0.001% success rate is still a business. There is no economic pressure to be good, only to be cheap.
Survivorship bias in your inbox
You only see what filters miss. AI-based filters already block the vast majority of spam. The sophisticated AI-written phish often gets through and does not look like spam at all, so you never count it as spam.
Attackers lag on purpose
Old kits keep working. Phishing is run by low-skill operators buying off-the-shelf kits. As long as returns are positive, upgrading tooling is optional.
AI raises the ceiling, not the floor
The real AI threat is the phish you cannot spot: flawless spear-phishing, cloned voices, fake invoices. The junk keeps flowing while the dangerous tier quietly gets better.
So the paradox dissolves
Obvious spam persists because it is profitable precisely by being obvious, and the best AI-crafted attacks are invisible to you by design.
4. Spot the Phish
Three emails, one is a modern AI-polished phishing attempt. Click (or press Enter on) the one you think is the attack.