cyber.hendryadrian.com
Security Insights · proofpoint.com

Anthropic Leak and Mercor Supply-Chain Attack: Lessons for Enterprise AI Security

· Original article ↗

Summary

The article uses the reported Anthropic source-code exposure and Mercor LiteLLM supply-chain compromise to highlight risks from release errors, dependencies and data flows, and urges stronger DLP, email security and visibility into AI use.

Key points

  • The article describes two Anthropic exposures, including Claude Code source code made public after a release-packaging error and separate publicly accessible internal files.
  • It reports that malicious code in the LiteLLM dependency exposed data flowing through Mercor’s AI systems, including customer data, and harvested API keys and other data.
  • The article attributes the Mercor attack to Team PCP and Lapsus$, but provides no supporting technical evidence for that attribution.
  • It presents human error, third-party dependencies and limited visibility into AI workflows as central risks, rather than attacks on AI models themselves.
  • Its recommendations include monitoring data shared with AI tools, reviewing DLP controls, securing email and collaboration channels, and improving visibility into AI use.

Article Details

Topic
The Anthropic source-code leak and Mercor supply-chain compromise highlight risks from human error and compromised AI dependencies.

MITRE ATT&CK

T1195 · Supply Chain CompromiseMalicious code was embedded in the LiteLLM open-source dependency, enabling a supply-chain compromise.T1552.001 · Credentials In FilesCredential-stealing malware reportedly harvested API keys and other credentials from code, configurations, and data flows.

Threat Actors

LAPSUS$The article identifies the Lapsus$ hacking group as responsible for the Mercor supply-chain compromise involving malicious code in LiteLLM.TeamPCPThe article identifies Team PCP as responsible for the Mercor supply-chain compromise involving malicious code in LiteLLM.

Vendors

AnthropicRecent incidents — the Anthropic source-code leak and the Mercor supply-chain compromise — demonstrate that AI security failures are occurring now and expose sensitive data, internal systems, and proprietary technology.

Products

Claude CodeThe Anthropic incident resulted from a release packaging error that publicly exposed internal files and Claude Code source code.Microsoft 365 Apps for EnterpriseEnterprises must prioritize preventing human-error and supply-chain exposures across AI integrations, dependencies like LiteLLM, and cloud ecosystems such as Microsoft 365.

Related Articles