All security briefings

Connected identities Security intrusion

A third-party AI tool became a route into connected systems

Incident
April 2026
Source report

What was reported

Vercel’s bulletin says the April incident began with a compromise of Context.ai, an AI tool used by an employee. The attacker took over the employee’s Google Workspace account, then reached the employee’s Vercel account and internal systems. Vercel reported exposure of environment variables not marked sensitive for a subset of customers. Its October 16 update states that remediation was complete and the incident closed.

Read the original report from Vercel

Our perspective

Our reading: a tool’s access can extend beyond the information someone deliberately pastes into it. Connected identities deserve the same scrutiny as selected records. This is relevant to Vivendur’s focus on explicit tool boundaries, alongside identity protection and vendor review.

Read about permissions between tools