Cloud development company Vercel disclosed unauthorized access to some internal systems and said a limited subset of customers was affected. The company engaged incident-response specialists, notified law enforcement and began working directly with affected customers, according to details reported by BleepingComputer from Vercel’s security updates.
Vercel said its services continued operating through the incident. It also said Next.js, Turbopack and its other open-source projects were not compromised. The company advised customers to examine environment variables, mark sensitive values appropriately and rotate secrets where necessary while the investigation continued.
An updated advisory attributed the entry point to a Google Workspace OAuth application associated with a third-party artificial-intelligence service. Vercel chief executive Guillermo Rauch said a company employee’s Google Workspace account was compromised through a breach at Context.ai. The intruder then moved from that account into Vercel environments and inspected environment variables that had not been designated sensitive.
Rauch said Vercel encrypts customer environment variables at rest, but its platform also allows variables to be classified as non-sensitive. Those values were not encrypted at rest, and the attacker gained further access by enumerating them, according to his account. Vercel subsequently added a dashboard overview for environment variables and improved controls for managing sensitive variables. Administrators and developers were urged to verify that secrets had not been placed in fields intended for non-sensitive configuration.
The disclosure followed a hacking-forum post from an actor who claimed affiliation with “ShinyHunters” and offered alleged Vercel data and access for sale. The poster claimed to hold source code, database material, access keys and credentials for internal deployments. The actor also shared a file said to contain 580 Vercel employee records and an image resembling an internal enterprise dashboard.
Those claims were not independently authenticated by BleepingComputer. People linked to other attacks attributed to the ShinyHunters extortion group also denied involvement in this incident, making the poster’s identity uncertain. Separately, the actor claimed in Telegram messages to have discussed a $2 million ransom with Vercel; the supplied reporting did not confirm that claim or say that Vercel was negotiating. The unverified forum material therefore provides context for the disclosure but cannot establish the breach’s full scope or the authenticity of specific data samples.
The incident highlights a distinction between encryption controls and the way customers classify configuration data. A value left in a non-sensitive category may become a route to broader access if it contains credentials or operational details. Vercel’s guidance therefore focused on reviewing classifications and rotating exposed secrets, rather than only checking whether its hosting services remained online.
