Mercor Confirms Cyber Attack: Lapsus$ Claims Theft of 4TB of Data

According to monitoring by 1M AI News, AI recruitment platform Mercor has confirmed it suffered a cyber attack due to a breach in the supply chain of the open-source Python library LiteLLM. Mercor stated it is “one of thousands of affected companies” and has hired third-party forensic experts to investigate. LiteLLM is a Python library that has been downloaded 97 million times in a month, used by developers as a unified interface to connect with over 100 AI services, including OpenAI and Anthropic. A hacker group named TeamPCP uploaded maliciously injected versions 1.82.7 and 1.82.8 to PyPI, which contained code that would steal SSH keys, API tokens, .env files, and cloud service credentials, while establishing a persistent backdoor. The malicious versions were removed within hours after being discovered by security company Snyk, but the exposure window was sufficient for attackers to infiltrate downstream systems. The ransomware hacker group Lapsus$ subsequently claimed responsibility for the attack on their leak site, asserting they stole approximately 4TB of data, including: 1. 939GB of source code 2. 211GB of databases 3. 3TB of storage buckets (allegedly containing video interview recordings, authentication documents, etc.) 4. All data from TailScale VPN. Lapsus$ also published some data samples in their post, including Slack communication records, ticketing system information, and videos of interactions between Mercor’s AI system and platform contractors. Security researchers on social media analyzed the leaked samples and noted the presence of internal project file structures suspected to be related to Amazon, Apple, and Meta, but Mercor has not confirmed which specific customer data was affected. Founded in 2023 with a valuation of $10 billion (Series C in October 2025), Mercor manages over 30,000 expert contractors and pays more than $2 million daily to contractors, providing expert-level human feedback services necessary for model training and evaluation for AI labs like OpenAI, Anthropic, and Google DeepMind. A Mercor spokesperson confirmed that an investigation has been initiated but declined to comment on whether the incident is related to Lapsus$'s claims, nor did they specify if any customer or contractor data was accessed, leaked, or misused. If Lapsus$‘s claims are true, this would represent a significant security incident directly impacting the core data of multiple leading AI labs’ training processes. The current relationship between TeamPCP and Lapsus$ remains unclear. Cybernews analysts suggest that Lapsus$'s attack on Mercor may signal a substantial collaboration between TeamPCP and the ransomware organization, similar to previous chain reactions following vulnerabilities exploited by ShinyHunters in Salesforce and Cl0p in MOVEit.