OpenAI announced a new "Trusted Contact" feature for ChatGPT on 7 May 2026, allowing users to designate a contact who can request account access if the user becomes unresponsive [source]. The feature is positioned as a safety mechanism for users who rely on ChatGPT to store sensitive information or manage critical workflows.

Under the system, a designated trusted contact can submit a request to OpenAI after a user-defined period of inactivity. OpenAI will then attempt to verify the user's status before granting access. The announcement does not specify what verification methods will be used, nor does it detail how OpenAI will confirm a user is genuinely incapacitated rather than simply inactive.

The feature raises questions about authentication boundaries and data custody. ChatGPT accounts may contain conversation histories, custom instructions, uploaded files, and API keys. Granting a third party access to this material—even a designated contact—introduces a new vector for social engineering, coercion, or unauthorised disclosure.

OpenAI states that users must explicitly enable the feature and that trusted contacts cannot access accounts without a formal request process. However, the announcement does not describe safeguards against abuse, such as mandatory cooling-off periods, multi-factor re-authentication, or audit logs visible to the original account holder.

The feature appears to be rolling out to ChatGPT Plus and Enterprise users first. OpenAI has not disclosed whether trusted contacts will be able to modify account settings, delete data, or export conversation logs. No independent security review of the feature has been published.

This development follows broader industry discussion about digital estate planning and post-mortem data access, but introduces the mechanism into an active AI service where account contents may include proprietary or confidential material.