AI Gateway Linked to Amazon Bedrock Hijacked for Cryptomining via Exposed SSH Access
What Happened — Darktrace reported that a LiteLLM AI gateway, which forwards requests to Amazon Bedrock, was compromised. Attackers leveraged exposed SSH access to install cryptomining software on the gateway’s underlying infrastructure.
Why It Matters for Compliance & Audit Readiness
- Unrestricted or poorly‑managed SSH access is a classic SOC 2 CC6.1 (Logical Access) failure, exposing organizations to unauthorized compute use and cost‑impact incidents.
- Continuous monitoring of privileged access and immutable audit logs provides the evidence needed to demonstrate “least‑privilege” controls during a SOC 2 audit.
- The incident underscores the need for documented SSH key rotation, MFA enforcement, and real‑time alerting—core elements of a defensible SOC 2 access‑control program.
Who Is Affected — SaaS AI‑gateway providers, cloud‑native developers, and any organization that integrates third‑party LLM APIs (e.g., fintech, health‑tech, media).
Recommended Actions
- Map the breach to SOC 2 CC6.1 and CC6.2 (System Operations) controls; verify that SSH keys are centrally managed, rotated, and protected with MFA.
- Deploy continuous privileged‑access monitoring and integrate alerts into your audit‑evidence repository.
- Conduct a rapid post‑incident review to confirm no data exfiltration occurred and update your incident‑response playbooks.
Source: HackRead – AI Gateway Connected to Amazon Bedrock Hijacked for Cryptomining
Technical Notes — The gateway’s SSH daemon was left reachable from the public internet, allowing attackers to gain root privileges and install cryptomining binaries. No specific CVE was cited; the weakness was a configuration/credential exposure. The compromised system was used to mine cryptocurrency, inflating cloud‑compute costs.