# Email threat library for AI agents and humans

> How email attacks work, from invoice fraud and phishing to prompt injection aimed at AI agents, and what stops each one before it is read.

Source: https://trotta.email/threats

Each attack, how it works, and what stops it at the door.

## Aimed at humans

### [Business email compromise (BEC)](https://trotta.email/threats/business-email-compromise)

What business email compromise is, its main forms, why a message with no link and no attachment passes filters, and the controls that stop it.

### [Invoice fraud by email](https://trotta.email/threats/invoice-fraud)

How a real-looking invoice with changed bank details reaches the finance team, the signs to check, and the controls that stop the payment.

### [CEO fraud](https://trotta.email/threats/ceo-fraud)

How CEO fraud uses an executive's name, urgency and secrecy to get a payment sent, the signs in the message, and the checks that stop it.

### [Payroll diversion](https://trotta.email/threats/payroll-diversion)

How a payroll diversion scam asks HR to change where a salary is paid, who is targeted, the signs, and the verification steps that stop it.

### [Vendor email compromise (VEC)](https://trotta.email/threats/vendor-email-compromise)

How vendor email compromise uses a real supplier's mailbox, or a close copy, to redirect a payment, the signs, and the controls that stop it.

### [Credential phishing](https://trotta.email/threats/credential-phishing)

How credential phishing leads from an email to a fake sign-in page, what it looks like now, why one-time codes get stolen too, and what stops it.

### [Spear phishing](https://trotta.email/threats/spear-phishing)

What spear phishing is, how it differs from mass phishing, why AI makes a tailored message cheap to write, the signs, and how to stop it.

## Delivery tricks

### [Lookalike domains](https://trotta.email/threats/lookalike-domains)

A lookalike domain is registered to be misread for a real one. How typosquatting and homograph attacks work in email, the signs, and what stops them.

### [Email spoofing](https://trotta.email/threats/email-spoofing)

Email spoofing is forging who a message says it is from. The three parts of a sender that can be faked, and what SPF, DKIM and DMARC do and miss.

### [Deceptive links](https://trotta.email/threats/deceptive-links)

A deceptive link shows one destination and opens another. How link text mismatches, redirects and shorteners work in email, the signs, and the fixes.

### [Malicious attachments](https://trotta.email/threats/malicious-attachments)

A malicious attachment is a file that carries malware or leads to it. The common carriers, why a PDF is no guarantee, the signs, and what stops them.

### [Disguised programs](https://trotta.email/threats/disguised-programs)

A disguised program is an executable named to look like a document, such as invoice.pdf.exe. How double extensions work, the signs, and what stops them.

### [QR code phishing](https://trotta.email/threats/qr-code-phishing)

QR code phishing, or quishing, hides a link in an image so the victim opens it on a phone. How it works, the signs, and the controls that stop it.

### [Thread hijacking](https://trotta.email/threats/thread-hijacking)

Thread hijacking inserts a malicious reply into a real email conversation. How reply chain attacks work, the signs, and how to verify before you act.

### [Account takeover](https://trotta.email/threats/account-takeover)

Account takeover is an attacker signing in as the real owner. How mailboxes are taken, what attackers do inside, and how codes and resets are the route in.

## Aimed at agents

### [Prompt injection by email](https://trotta.email/threats/prompt-injection-by-email)

Prompt injection by email is text written for the AI model that reads an inbox, not for the human. How the attack works, the signs, and what stops it.

### [Hidden text in email](https://trotta.email/threats/hidden-text-in-email)

Hidden text in email is invisible to a human and plain to an AI model. How white, tiny and off-screen text carries instructions, and how to stop it.

### [Hidden instructions in attachments](https://trotta.email/threats/hidden-instructions-in-attachments)

A PDF, document or spreadsheet can carry instructions for the AI agent asked to read it. How the attack works, the signs to look for, and how to stop it.

### [Agent hijacking](https://trotta.email/threats/agent-hijacking)

Agent hijacking turns an AI agent against its owner: it forwards, replies, pays or deletes on a stranger's word. How it works by email and how to stop it.

### [Data exfiltration through agents](https://trotta.email/threats/data-exfiltration-through-agents)

Data exfiltration through agents gets an AI agent to send out what it can reach: secrets, files, other messages. How it works by email and how to stop it.

### [Verification code theft](https://trotta.email/threats/verification-code-theft)

Verification code theft uses an AI agent with mailbox access to read a human's sign-in codes and reset links. How the takeover works and how to stop it.
