An indirect prompt injection flaw in GitLab's artificial intelligence (AI) assistant could have allowed attackers to steal source code, direct victims to malicious websites, and more. In fact, ...
Varonis found a way to chain three bugs into one exploit that can lead to data exfiltration.