Cloud infrastructure provider Vercel disclosed over the weekend that its internal networks were infiltrated by cybercriminals, resulting in unauthorized access to client information. The attackers have asserted they exfiltrated confidential customer credentials and are attempting to offload this information on dark web marketplaces.
According to a Sunday announcement, the initial point of entry was traced to Context AI, a separate software developer. A Vercel staff member installed an application developed by Context AI and linked it to their Google-hosted corporate account. The threat actors exploited this OAuth integration to hijack the employee’s Google account, subsequently penetrating certain Vercel internal networks and accessing unencrypted credentials.
Vercel confirmed that its core open-source initiatives, including Next.js and Turbopack, remain secure and unaffected by this incident. These tools are extensively utilized by web and application developers globally.
The company has reached out to affected clients whose application data and authentication keys were compromised.
In a message on X, Vercel CEO Guillermo Rauch urged users to refresh any keys and credentials labeled as “non-sensitive” within their application deployments.
The identity of the perpetrators behind the Vercel and Context AI breaches remains unknown, as does whether the same group is responsible for both. The individual marketing the stolen information online identified themselves as part of the ShinyHunters hacking collective. A listing viewed by TechCrunch alleged that the group is offering access to customer API keys, source code repositories, and database records extracted from Vercel.
However, the ShinyHunters group, previously associated with attacks on cloud-based and database firms, explicitly denied involvement to cybersecurity outlet Bleeping Computer.
As details of the intrusion continue to surface, this event adds to a recent trend of “supply chain” attacks targeting software developers whose code underpins much of the internet. By compromising widely adopted software supporting web infrastructure, attackers can simultaneously harvest credentials from numerous targets and expand access to vast datasets held by other major cloud providers.
Vercel provided limited additional details, stating it is actively investigating and has requested information from Context AI. The company warned the breach could impact “hundreds of users across many organizations,” suggesting potential cascading security issues throughout the technology sector.
Context AI, which develops evaluation and analytics tools for AI models, acknowledged on its site a March breach involving its Context AI Office Suite consumer application. This app enables users to automate tasks across various third-party services via an undisclosed third-party provider.
Context AI stated it initially notified one customer about the breach but now suspects the scope is wider than originally believed, based on Vercel’s findings. The company noted that hackers “likely compromised OAuth tokens for some of our consumer users.”
Context AI did not respond to requests for comment or inquiries regarding the breach. It remains unclear why Context AI did not disclose the incident immediately or if the company faced extortion demands, such as ransom.
Vercel also declined to answer questions about the incident, including how many customers might be impacted.
Corrected to remove a reference to an unrelated Context AI whose staff were acquired by OpenAI.
Topics