The Phishing Attacks Slipping Past Your Email Filters

Email security gateways have become very good at stopping the obvious attacks: known malicious attachments, spoofed domains and links to sites already flagged as dangerous. Attackers know this. So they’ve changed tactics, designing campaigns specifically to look clean to automated filters while still fooling the people who read them.

The result is a new generation of phishing that lands in inboxes looking legitimate. Here are the techniques security teams are seeing most, and how to defend against them.

1. Abusing trusted services

Instead of linking to a suspicious domain, attackers host phishing pages or malicious files on platforms your filters trust: file-sharing services, e-signature tools, online form builders, collaboration sites and cloud storage. The email often comes from the legitimate service itself, with a valid sender reputation and a link to a real domain. The malicious content sits one click deeper.

2. QR code phishing

A QR code embedded in an image or PDF contains no clickable link for filters to inspect. When users scan it with a phone, they leave the protected corporate environment entirely, often landing on a convincing login page on a device with fewer security controls. Fake MFA reset notices and document-sharing alerts are common lures.

3. Adversary-in-the-middle kits

Modern phishing kits sit between the user and the real login page, relaying credentials and MFA codes in real time and capturing the session token. The user sees a legitimate sign-in experience, completes MFA and never realizes the attacker now has an authenticated session.

4. Compromised vendor and partner accounts

When attackers take over a real supplier’s mailbox, their emails come from a trusted domain with passing authentication checks and an existing conversation history. A request to update bank details on an upcoming invoice, replying in an active thread, is one of the hardest attacks for any filter to catch.

5. AI-crafted lures

Generative AI lets attackers write fluent, personalized messages at scale, eliminating the spelling errors and awkward phrasing users were trained to spot. Messages can reference real projects, colleagues and recent events pulled from public sources.

6. Callback and multichannel attacks

Some emails contain no link or attachment at all, just a fake invoice or subscription renewal with a phone number to call. The attack happens on the phone, where a fraudster talks the victim into installing remote access software or sharing credentials. Others move the conversation to text messages, Teams or Slack, outside email security entirely.

How to defend

Adopt phishing-resistant MFA. Passkeys and FIDO2 security keys stop adversary-in-the-middle credential theft, because they won’t authenticate to a fake site.

Add behavioral and AI-based email analysis. Tools that model normal communication patterns can flag unusual requests, first-time senders, tone shifts and payment changes that signature-based filters miss.

Protect sessions. Monitor for suspicious session reuse, shorten token lifetimes for sensitive applications and revoke sessions quickly when compromise is suspected.

Verify payment changes out of band. Require a callback to a known phone number before changing vendor banking details or approving unusual transfers.

Extend protection beyond email. Apply security controls to collaboration platforms and mobile devices, where phishing increasingly arrives.

Update training. Teach users about QR codes, callback scams, polished AI-written messages and requests that arrive through trusted services. Make reporting easy and reward it.

The bottom line

Phishing hasn’t been solved. It has adapted to the defenses. Organizations that pair modern email analysis with phishing-resistant authentication, strong payment controls and realistic user training will catch the attacks that filters alone let through.