Gist: The post argues that prompt injection remains a major AI security risk and says the safest choice is to avoid granting AI access to email. It frames this as a practical precaution rather than a feature critique.
Signal reason: It reinforces a broader positioning around AI safety and cautious data access.
