ML

How it works

• The system is shown many examples and builds its own rules for telling them apart.
• It powers spam filters, fraud checks and unusual-login detection.
• It adapts to new tricks that fixed rules miss.
• It can be wrong, so important decisions still get a human check.

What can go wrong (example scenario)

A five-person retailer relied on a keyword-based spam filter. Attackers simply avoided the keywords. A plain, polite email asking for a “quick update to the supplier's banking details” sailed through, because no rule had been written for politeness.

How WILDCARD handles it

The machine learning that helps you sits inside other cards: Email Security scores messages, Workstation Defender scores behaviour, and Identity Monitor spots logins that break your normal pattern.

• A dental office's email filter learns that invoices from its real suppliers look a certain way and flags imposters.
• A law firm's identity monitoring learns each lawyer's usual hours and locations.
• A retailer's card processor flags a transaction that doesn't fit the shop's pattern.

Example scenario: A small clinic's receptionist received an email from a known supplier's real address with slightly off phrasing and a new bank account. The filter's model scored it as likely compromise, based on language and timing, and held it. The supplier confirmed by phone that their mailbox had been hijacked.

A question owners ask

Is machine learning the same as AI?

ML is one way of building AI, the way that learns from data. When a security tool says it uses AI, it almost always means ML models scoring what they see.