Techniques are described for automatically distributing validated user safety alerts from a networked computing device. The networked computing device may be configured to operate as an autonomous agent to perform actions on behalf of a user without receiving direct instructions from the user. For example, the autonomous agent computing device may be configured to make certain purchases, send alerts or reminders, or perform other functions in accordance with preprogrammed rules. According to the disclosed techniques, the autonomous agent computing device is configured to automatically generate and send an alert to one or more computing devices associated with the user upon detecting a safety concern for the user. The autonomous agent also uses a signing key associated with its digital certificate, which verifies the identity of the autonomous agent, to sign the alert such that a third-party server may validate the alert prior to distribution to the destination computing devices.