Vercel Says Breach Began with Compromised Context.ai OAuth Access
Summary
Vercel says a compromise of a third-party AI tool’s Google Workspace OAuth app enabled attackers to take over an employee account and access some environments; a limited subset of customers had credentials or environment variables compromised.
Key points
- Vercel says the incident began with a compromise of a third-party AI tool’s Google Workspace OAuth app, which the article identifies as Context.ai.
- Attackers used that access to take over a Vercel employee’s Google Workspace account and access some internal environments and non-sensitive environment variables.
- Vercel says a limited subset of customers had credentials or environment variables compromised and were notified directly.
- Affected customers were advised to rotate credentials, environment variables and Deployment Protection tokens, and review logs and deployments for suspicious activity.
- Vercel engaged Google Mandiant and other experts, notified law enforcement, and added protections while its investigation continued.
- Context.ai reported unauthorized access to its AWS environment and said the attacker likely compromised some consumer OAuth tokens; it also said a compromised token appeared to have been used to access Vercel’s workspace.
- A BreachForums post claimed the breach for ShinyHunters, but the post was removed and the group denied involvement.
Article Details
- Victim Organization
- Vercel
- Incident Type
- Third-party AI tool compromise led to misuse of a Google Workspace OAuth token, takeover of a Vercel employee account, and access to some Vercel environments and non-sensitive environment variables.
- Data Types Exposed
- Credentials
- Environment variables not marked as sensitive
- Google Workspace OAuth tokens
- Affected Records
- A limited subset of Vercel customers; exact number not disclosed.
- Operational Impact
- Attackers accessed some internal systems and environments. Vercel reported that credentials or environment variables belonging to a limited subset of customers were compromised. Affected customers were notified and advised to rotate credentials, environment variables, and Deployment Protection tokens, and review activity logs and deployments. No service availability impact is reported.
- Ransom Or Extortion
- A BreachForums post claiming to be from ShinyHunters, or from someone impersonating the group, attempted to sell information allegedly stolen in the breach. The post was removed, and ShinyHunters denied involvement. The article does not confirm that the group stole or sold the information.
- Claim Status
- confirmed
MITRE ATT&CK
People
Guillermo RauchVercel CEO who provided statements about the breach, customer environment-variable protections, and his suspicion that the attackers were highly sophisticated and accelerated by AI.Jaime BlascoNudge Security CTO who linked the third-party AI tool to Context.ai based on a removed Chrome extension listing associated with the OAuth grant.
Threat Actors
Vendors
Context.aiVercel suffered a security breach after a third-party AI tool compromise (Context.ai) allowed attackers to hijack a Vercel employee’s Google Workspace account and access some internal environments and environmentGoogleafter a third-party AI tool compromise (Context.ai) allowed attackers to hijack a Vercel employee’s Google Workspace account and access some internal environments and environment variables.MandiantA limited subset of customers had credentials compromised, and Vercel has notified affected users, advised rotating secrets, engaged Google Mandiant and other experts, and deployed additional protections.VercelVercel suffered a security breach after a third-party AI tool compromise (Context.ai) allowed attackers to hijack a Vercel employee’s Google Workspace account and access some internal environments and environment
Products
Context.aiVercel suffered a security breach after a third-party AI tool compromise (Context.ai) allowed attackers to hijack a Vercel employee’s Google Workspace account and access some internal environments and environmentGoogle Workspaceafter a third-party AI tool compromise (Context.ai) allowed attackers to hijack a Vercel employee’s Google Workspace account and access some internal environments and environment variables.Next.jsVercel engaged Google Mandiant and other experts, notified law enforcement, deployed additional protections, and audited its supply chain including Next.js and Turbopack.Turbopack#Vercel