ChatGPT for Google Sheets vulnerable to data exfiltration via prompt injection

ChatGPT for Google Sheets is vulnerable to data exfiltration and phishing overlay attacks that affect workbooks across the victim's account after an indirect prompt injection in a single sheet. This attack does not require human-in-the-loop approvals, even when settings require human approval before ChatGPT edits workbooks. A single indirect prompt injection triggered The vulnerability occurs when any untrusted data source manipulates ChatGPT to run an attacker-controlled external script leveraging permissions granted to the extension. The issue was responsibly disclosed to OpenAI with no substantive response beyond an automated reply. OpenAI documentation does not describe the sensitive capabilities or risks of indirect prompt injection.