# Amazon SES Weaponized for Phishing & BEC via Leaked AWS IAM Access Keys (Securelist, May 2026)

> Kaspersky/Securelist documents a 2026 trend in which threat actors harvest leaked AWS IAM access keys from public GitHub repos, ENV files, Docker images, configuration backups, and open S3 buckets — primarily via TruffleHog-based bots — and then weaponize compromised Amazon SES tenants. The hijacked SES tenants emit fully SPF/DKIM/DMARC-authenticated phishing and BEC emails carrying .amazonses.com Message-ID headers and amazonaws.com-hosted phishing landing pages, defeating reputation-based controls and AP/finance review. Observed lures include fake DocuSign signature requests and fabricated employee-vendor invoice threads driving urgent wire transfers.

- **Published:** 2026-05-04T12:00:00Z
- **Last reviewed:** 2026-05-04T12:00:00Z
- **Canonical:** https://intel.threadlinqs.com/threat/TL-2026-0451
- **Card image:** https://intel.threadlinqs.com/og/threat/threat-intel-amazon-ses-leaked-key-bec-tl-2026-0451.webp?v=a13d1e50 (Threadlinqs threat intel: Amazon SES Leaked-Key BEC Phishing: Detection & IOCs (TL-2026-0451) — High severity, 22 IOCs, 9 detection rules.)
- **ID:** TL-2026-0451
- **Severity:** HIGH
- **Category:** PHISHING
- **Status:** ACTIVE
- **Detections:** 9 (1 shown free below) · **IOCs:** 22 (2 shown free below)

## Description

OVERVIEW
In early 2026, Kaspersky's Roman Dedenok (Securelist, 2026-05-04) reported a sustained uptick in phishing and Business Email Compromise (BEC) campaigns abusing Amazon Simple Email Service (Amazon SES). Unlike commodity phishing that depends on freshly registered look-alike domains and quickly blocklisted infrastructure, this trend hinges on hijacking legitimate AWS tenants. The attacker piggybacks on Amazon's reputation, authentication footprint, and IP space. Because emission occurs from real AWS-owned IPs, with valid SPF/DKIM signatures and aligned DMARC, the messages bypass reputation, header-based, and protocol-level email defenses that secure operations centers traditionally rely upon.

INITIAL ACCESS — IAM ACCESS KEY HARVESTING
The campaign begins with credential acquisition rather than exploitation. Phisher infrastructure runs automated scrapers built around the open-source secret-detection utility TruffleHog and similar tools, continuously enumerating public GitHub repositories, mirror sites (e.g., GitHub commit history, gist.github.com), public S3 buckets, leaked Docker image layers (via Docker Hub and self-hosted registries), and exposed environment files (.env, config.yml, docker-compose.yml). Long-lived AWS IAM access keys, especially with patterns matching `AKIA[0-9A-Z]{16}`, are tested against AWS APIs to determine permission scope and SES sending limits (`GetSendQuota`, `ListIdentities`, `GetAccount`). Keys with attached `AmazonSESFullAccess`, `ses:SendEmail`, or overly permissive admin policies are retained for weaponization.

WEAPONIZATION — SES TENANT ABUSE
Once a valid key is identified, attackers use the AWS SDK or smtp credentials derived from the IAM key (Amazon SES SMTP credentials are deterministically derived from the IAM secret access key) to issue `SendEmail`, `SendBulkTemplatedEmail`, and `SendRawEmail` API calls. Custom HTML templates supplied via SES enable convincing brand impersonation. Because the legitimate tenant is already verified for one or more sender domains and has DKIM signing keys provisioned in Route53 (or equivalent DNS), every outbound message carries valid DKIM-Signature, an `X-SES-Outgoing` header, and a Message-ID of the form `<UUID@email.amazonses.com>` or `<UUID@*.amazonses.com>`. SPF passes via Amazon's published `_spf.amazonses.com` include directives. Provided the tenant's DMARC record is `p=none`/`p=quarantine`, the abusing email aligns and passes DMARC.

LURE CATALOG — DOCUSIGN AND BEC
In observed campaigns, two lure categories dominate: (1) Fake DocuSign / electronic-signature notifications inviting the victim to review and sign a document. The link in the email is masked as an `amazonaws.com` URL — typically a public S3 bucket or an `*.s3.amazonaws.com` static site hosting the phishing form — which redirects to or directly serves a credential-harvesting page styled after Microsoft 365, Google Workspace, or DocuSign sign-in. (2) BEC variants in which the attacker fabricates an entire reply chain between an internal employee and an external service provider regarding an outstanding invoice. The body contains the forwarded thread; PDF attachments include payment details and forged supporting documentation. Crucially, the PDFs in observed BEC samples carry no malicious URLs or QR codes — they are passive financial fraud documents intended to anchor the social-engineering ask. The email is addressed to finance / accounts payable, requesting urgent wire transfer to an attacker-controlled bank account.

WHY EMAIL CONTROLS FAIL
Standard defensive layers — sender reputation, IP blocklists, SPF/DKIM/DMARC enforcement, and even some BEC-tuned ML detectors that flag domain-anomaly senders — under-detect this campaign because: (a) the sending IP belongs to AWS (e.g., 23.249.208.0/20, 67.231.144.0/20, 199.122.120.0/22 SES ranges) and cannot safely be blocklisted; (b) authentication results report `dkim=pass header.d=<victim-tenant-domain>` and `spf=pass`; (c) Message-ID `.amazonses.com` is widely associated with legitimate transactional mail, so its presence is a positive trust signal in many SOC playbooks; (d) outbound link clicks resolve to `amazonaws.com` which most secure web gateways categorize as `cloud-services` or `business-and-economy`. The attacker is, in effect, borrowing the victim tenant's earned reputation.

DEFENDER LEVERAGE POINTS
The defensive opportunity sits at four control planes: AWS-side (IAM hygiene, key rotation, MFA, IP-restricted policies, CloudTrail SES anomaly detection, GuardDuty `Recon:IAMUser/MaliciousIPCaller` and `Stealth:IAMUser` findings); developer pipeline (pre-commit secret scanners, GitHub push-protection, Docker image scanning); inbound email (header-level inspection of `X-SES-Outgoing`, sudden tenant-domain change in `dkim=pass header.d`, link-rewriting/sandboxing of amazonaws.com links, attachment defanging); and finance process (out-of-band verification for any wire-transfer change, especially when the request originates from an internal employee impersonation). For sending tenants, Amazon's `ListIdentities`, `GetSendStatistics`, `GetSendQuota`, and SES Event Publishing (CloudWatch / SNS bounces, complaints, deliveries) provide ground truth on whether their tenant is being abused.

ATTRIBUTION
Kaspersky/Securelist did not attribute the activity to a named actor. The tradecraft (mass IAM key harvesting + SES abuse + Docusign-themed lures + BEC) overlaps with multiple financially motivated clusters historically tracked under labels such as TA4903, TA4557, and various BEC actors. Attribution is therefore listed as Unknown with LOW confidence pending hash, infrastructure, or actor-overlap evidence.

## MITRE ATT&CK

- T1593 Search Open Websites/Domains
- T1593.003 Search Open Websites/Domains: Code Repositories
- T1596 Search Open Technical Databases
- T1586 Compromise Accounts
- T1586.003 Compromise Accounts: Cloud Accounts
- T1583.006 Acquire Infrastructure: Web Services
- T1585.002 Establish Accounts: Email Accounts
- T1608.002 Stage Capabilities: Upload Tool
- T1078.004 Valid Accounts: Cloud Accounts
- T1566 Phishing
- T1566.001 Phishing: Spearphishing Attachment
- T1566.002 Phishing: Spearphishing Link
- T1552.001 Unsecured Credentials: Credentials In Files
- T1552.004 Unsecured Credentials: Private Keys
- T1552.005 Unsecured Credentials: Cloud Instance Metadata API
- T1684.001 Impersonation
- T1036 Masquerading
- T1098.001 Account Manipulation: Additional Cloud Credentials
- T1526 Cloud Service Discovery
- T1580 Cloud Infrastructure Discovery
- T1534 Internal Spearphishing
- T1213.003 Data from Information Repositories: Code Repositories
- T1102 Web Service
- T1657 Financial Theft
- T1496 Resource Hijacking

## Free detection (Splunk SPL, Microsoft KQL, Sigma)

**Inbound Amazon SES Email With amazonaws.com Link Payload — Exploitation Phase** (`TL-2026-0451-SPL-001`)

Detects inbound email that carries Amazon SES outbound headers (X-SES-Outgoing or .amazonses.com Message-ID) AND a body link to amazonaws.com / *.s3.amazonaws.com. This co-occurrence is a high-fidelity signal for the TL-2026-0451 SES phishing/BEC campaign in which attackers weaponize compromised SES tenants and host credential-harvest pages on AWS-owned infrastructure.

- **Languages:** Splunk SPL, Microsoft KQL, Sigma
- **ATT&CK:** T1566.002, T1078.004, T1656
- **Data sources:** email_gateway, ms_defender_for_office_365, proofpoint, mimecast, cisco_esa
- **Severity:** high
- **Confidence:** high

### Splunk SPL

```spl
index=email sourcetype IN ("ms:o365:reporting:messagetrace","proofpoint:smtp","mimecast:receipt","cisco:esa:textmail","email_gateway") earliest=-24h
| eval has_ses_outgoing=if(match(_raw, "(?i)X-SES-Outgoing:"), 1, 0)
| eval has_ses_msgid=if(match(message_id, "(?i)<[A-Za-z0-9-]+@.*\.amazonses\.com>"), 1, 0)
| eval has_aws_link=if(match(body, "(?i)https?://([a-z0-9.-]+\.)?(s3\.amazonaws\.com|amazonaws\.com)/"), 1, 0)
| where (has_ses_outgoing=1 OR has_ses_msgid=1) AND has_aws_link=1
| stats count, values(sender) AS senders, values(recipient) AS recipients, values(subject) AS subjects, values(message_id) AS msg_ids by src_ip, header_d
| where count >= 1
| table _time, src_ip, header_d, senders, recipients, subjects, msg_ids, count
```

### Microsoft KQL

```kql
EmailEvents
| where Timestamp > ago(24h)
| where EmailDirection == "Inbound"
| join kind=inner (EmailUrlInfo | where Timestamp > ago(24h)) on NetworkMessageId
| extend HasSesHeader = AuthenticationDetails has "amazonses.com" or AdditionalFields has "X-SES-Outgoing"
| extend HasAwsLink = Url matches regex @"(?i)https?://([a-z0-9.-]+\.)?(s3\.amazonaws\.com|amazonaws\.com)/"
| where HasSesHeader and HasAwsLink
| project Timestamp, NetworkMessageId, SenderFromAddress, SenderMailFromDomain, RecipientEmailAddress, Subject, Url, AuthenticationDetails
| sort by Timestamp desc
```

### Sigma

```yaml
title: Inbound Amazon SES Email With amazonaws.com Link Payload
id: a4510001-1111-4451-9001-aaaaaaaaaaaa
status: experimental
description: Detects inbound email that carries Amazon SES outbound headers (X-SES-Outgoing or .amazonses.com Message-ID) and a body link to amazonaws.com or s3.amazonaws.com. High-fidelity signal for the TL-2026-0451 SES phishing/BEC campaign.
references:
    - https://securelist.com/amazon-ses-phishing-and-bec-attacks/119623/
    - https://attack.mitre.org/techniques/T1566/002/
    - https://attack.mitre.org/techniques/T1078/004/
author: Threadlinqs Team
date: 2026/05/04
tags:
    - attack.initial_access
    - attack.t1566.002
    - attack.t1078.004
logsource:
    product: email_gateway
    category: mail
detection:
    selection_header:
        headers|contains: 'X-SES-Outgoing:'
    selection_msgid:
        message_id|re: '<[A-Za-z0-9-]+@.*\.amazonses\.com>'
    selection_link:
        body|re: '(?i)https?://([a-z0-9.-]+\.)?(s3\.amazonaws\.com|amazonaws\.com)/'
    condition: (selection_header or selection_msgid) and selection_link
falsepositives:
    - Legitimate transactional mail from a known SES tenant that links to a public AWS-hosted asset (e.g., release notes, status pages)
    - Internal notifications from SaaS vendors that legitimately host static assets in S3
level: high

```

**Known false positives:**

- Legitimate transactional mail from a known SES tenant linking to a public AWS-hosted asset (release notes, status pages)
- Internal notifications from SaaS vendors that legitimately host static assets in S3 (e.g., signed-URL downloads)
- Marketing emails from vendors using SES to deliver newsletters with CDN links served from amazonaws.com

**Tuning notes:** Allowlist header.d values for known-trusted SES tenants (vendor newsletters, monitoring providers) to reduce noise. Tighten the link regex to exclude allowed S3 buckets (e.g., your CI artifact bucket, vendor signed URLs). Consider adding a body keyword filter (docusign, signature, invoice, wire, payment) to elevate to critical when the lure language matches BEC/DocuSign tradecraft.

### 8 more detection rules (Splunk SPL, Microsoft KQL, Sigma), unlock with TL Creds

- Defender for Office 365 — Inbound SES Phishing With [indicator] Link — Exploitation Phase (locked: unlock with TL Creds)
- Sentinel — IAM User SES Recon Burst Followed by SendEmail — Post-Exploitation Phase (locked: unlock with TL Creds)
- Sentinel — Anomalous Outbound SES SendEmail Volume Spike — C2 / Mass-Mail Phase (locked: unlock with TL Creds)
- AKIA AWS Access Key Pattern in Source Artifact — Exploitation Phase (locked: unlock with TL Creds)
- Additional IAM Access Key Created Followed by SES API Use — Post-Exploitation / Persistence Phase (locked: unlock with TL Creds)
- Outbound SES SMTP Connection From Non-Production Host — C2 / Lateral Movement Phase (locked: unlock with TL Creds)
- AWS CloudTrail SES Discovery Burst From IAM User — Post-Exploitation Phase (locked: unlock with TL Creds)
- AWS SES SendEmail / SendRawEmail Abuse From New Source IP — C2 / Mass-Mail Phase (locked: unlock with TL Creds)

Sign up free to unlock the detection rules: https://threadlinqs.com/?modal=signup

## Sample indicators (2 of 22)

Defanged, TLP:CLEAR.

- Domain: `amazonses[.]com`
- Domain: `amazonaws[.]com`

Full set of 22 indicators unlocks with TL Creds: https://threadlinqs.com/tl-creds

## Sources

- ["Legitimate" phishing: how attackers weaponize Amazon SES to bypass email security](https://securelist.com/amazon-ses-phishing-and-bec-attacks/119623/)
- [Amazon SES Developer Guide — Authentication and Identity Management](https://docs.aws.amazon.com/ses/latest/dg/send-email-authentication.html)
- [Identity-based policies for Amazon SES (least-privilege examples)](https://docs.aws.amazon.com/ses/latest/dg/sending-authorization.html)
- [AWS Security Best Practices — IAM Access Keys](https://docs.aws.amazon.com/IAM/latest/UserGuide/best-practices.html)
- [GuardDuty finding types — IAM and credential-access related](https://docs.aws.amazon.com/guardduty/latest/ug/guardduty_finding-types-iam.html)
- [TruffleHog — Find leaked credentials](https://github.com/trufflesecurity/trufflehog)
- [MITRE ATT&CK — T1078.004 Valid Accounts: Cloud Accounts](https://attack.mitre.org/techniques/T1078/004/)
- [MITRE ATT&CK — T1552.001 Unsecured Credentials: Credentials In Files](https://attack.mitre.org/techniques/T1552/001/)
- [MITRE ATT&CK — T1566 Phishing](https://attack.mitre.org/techniques/T1566/)
- [MITRE ATT&CK — T1657 Financial Theft](https://attack.mitre.org/techniques/T1657/)
- [FBI IC3 — Business Email Compromise (BEC) Public Service Announcement](https://www.ic3.gov/Media/Y2024/PSA240911)
- [CISA — Identifying and Mitigating Living-Off-the-Cloud Tradecraft](https://www.cisa.gov/news-events/cybersecurity-advisories)
- [Amazon SES SMTP credential derivation from IAM secret access key](https://docs.aws.amazon.com/ses/latest/dg/smtp-credentials.html)

## Full data

The other 8 detection rules (Splunk SPL / Microsoft KQL / Sigma) unlock with TL Creds; a free account includes daily creds: https://threadlinqs.com/?modal=signup
Indicator values unlock with TL Creds, at a lower price on Red and Purple; AI agents can query rules and IOCs through the Threadlinqs MCP server (Purple): https://intel.threadlinqs.com/mcp

Canonical: https://intel.threadlinqs.com/threat/TL-2026-0451
