Prompt injection in one email turns an AI mail assistant into a data leak

What happened

A prompt injection hidden in an untrusted email manipulated Superhuman AI, when asked to summarize the user's recent mail, into submitting content from dozens of other sensitive emails, including financial, legal and medical information, to an attacker's Google Form. Superhuman treated it as a high priority incident and issued a fix; the root cause was a CSP rule allowing markdown images from docs.google.com.