An online platform that matches people at risk of being detained with peers all over the world who are able to trigger appropriate actions when they stop receiving "I'm fine" notifications from that person. Advantages of such direct person to person links would be:
a) guardians are involved over a longer period of time and are therefore familiar with the context and risks of the person concerned, b) the person at risk has a real, personal social peer which might make for extra emotional support when anonymity and being forgotten is the biggest fear c) it does not rely on a stress signal being sent which might be impossible
and d) any means of communications (twitter/email/sms...) could be possible, depending of the case
The core concept is to personalise stress signals through a network of guardians for people at risk of being detained. This personalisation provides support to the person at risk, but also allows the guardian to become familiar with a particular case and better judge what the appropriate actions are when they stop receiving OK messages.
There is also another important aspect related to raising awareness. I believe, humans easily loose interest in things that are big and anonymous. So, donating to AI to pay for human-rights lawyers is important, but does not engage the general public in a deeper way. However, when people are agreeing to be a guardian and monitor the well-being of a certain person at risk, there is a real sense of personal responsibility and people will devote time and effort to learn about the circumstance which brought the individual in this situation which further improves the understanding of the broader public of the problems related to human-rights abuse.
Of course, certain mechanism have to be in place that ensures that not a single guardian being temporarily unavailable puts the individual at risk. So, for example, there could be a team of guardians for each case which communicate over the online platform and check with each other about the status.