In CI/CD pipelines, the weakest link is often not the code, not the infrastructure, but the human. Social engineering attacks on CI/CD are rising fast. Phishing, fake commits, forged build notifications — these are precision strikes aimed at breaking trust in automated delivery. One mistaken approval in a pull request, one leaked token in a chat, one click on a fake merge alert, and the attacker owns the build.
Continuous Integration and Continuous Delivery rest on trust. You trust that the code being merged is reviewed. You trust that your pipeline files are untouched. You trust that your build server runs only what you approved. Social engineering targets this trust layer. It pushes people to bypass controls without noticing.
The risk is bigger when pipelines trigger deployments automatically. If an attacker tricks a developer into updating a dependency from a malicious source, the pipeline will run it. The deployment will succeed. The backdoor is now live. No alarms go off because nothing technically failed. From the system’s view, it was a valid build triggered by an authorized user.