Startup Mercor hit through a compromise of open-source project LiteLLM

AI recruiting startup Mercor has confirmed a cyberattack and data breach after an extortion-focused hacking group claimed responsibility for breaking into the company's systems. The attack vector is particularly worrying: the attackers exploited a compromise of the popular open-source project LiteLLM, a tool that a significant share of modern AI infrastructure relies on.

LiteLLM is a widely used open-source library that acts as a universal proxy gateway for working with language models from different providers: OpenAI, Anthropic, Google, Mistral, Cohere and dozens of others. The tool became popular among tech startups and corporate AI teams because it lets them unify calls to different LLMs through a single API, manage key rotation and keep centralized logs of requests. That position in the infrastructure, between the application and the language models, with access to API keys and user data, makes LiteLLM an especially attractive target for attackers.

The compromise of such a component goes far beyond an incident at a single company. It is a supply chain attack: any project using the vulnerable version of LiteLLM was potentially at risk. As of publication, it remains unknown how many other organizations may have been affected by the same compromise.

Mercor is a US-based AI startup that specializes in recruiting technical specialists using artificial intelligence. The platform uses language models to automatically assess candidates' skills, conduct structured interviews and match specialists with open positions at AI-focused companies. By the nature of its business, Mercor's systems hold highly sensitive information: résumés and professional profiles of thousands of job seekers, technical assessment results, salary expectation data, and details about employer companies and their hiring needs.

The group that claimed responsibility follows a long-established double extortion model: the victim receives an ultimatum to pay a ransom or lose control over the stolen information. Unlike classic ransomware attacks, where a company loses access to its own files, this scheme exploits fear of regulatory penalties and reputational damage. For a platform handling candidates' personal data, a public leak can trigger regulatory investigations and an irreversible loss of user trust.

Mercor has confirmed the incident, but the company has not yet disclosed the scale of the stolen data or the attackers' demands.

The incident exposes a systemic problem in the modern AI sector: hundreds of companies depend on the same set of open-source tools, including LiteLLM, LangChain, LlamaIndex and Transformers. Reusing a common stack speeds up development but concentrates risk: compromising one popular component hits that component's entire user ecosystem at once. Attacks on open-source projects are becoming more targeted: attackers analyze the dependency graph, study a library's user ecosystem and pick the entry point with the widest victim reach.

For companies building AI products on open-source proxy tools with access to API keys and user data, the Mercor case is a direct warning: regular dependency audits, monitoring of security updates and isolation of privileged infrastructure components should become standard practice, not an item on a perpetually postponed to-do list.