Facebook mental-health safety: features for users showing signs of depression or self-harm risk

Public debate over the past few years has focused on how social media can deepen feelings of depression and isolation. On Facebook, many people also reveal in their messages or comments that they are depressed, and in many cases they signal that they intend to take their own life or hurt themselves. As a product manager at Meta, what features would you build into the core Facebook app to combat depression, suicide and self-harm? Explain whom each feature serves, how it works, how you would know it is helping and what could go wrong. ```hint Match the response to the risk Someone signaling that they are in immediate danger, someone who has felt low for months and a friend who notices a worrying post need different things, and getting each one wrong has a different cost. Decide which of them you design for first. ``` ### Clarifying Questions - Is the scope only people who already show signs of distress, or also the ways Facebook itself may contribute to depression and isolation? - Which signals may the product use: public posts and comments only, or private messages too, which carry stricter privacy expectations and may be end-to-end encrypted? - Will Meta work with crisis lines and mental-health organizations, and in which countries would this launch first? - How should the product treat teenagers, who may need different protections and escalation paths? ### What a Strong Answer Covers - A goal framed around safety and wellbeing rather than engagement, including a position on Facebook's own contribution to the problem - Segmentation by level of risk and by role (the person struggling, friends and family who notice, the wider user base), with a justified focus - Features that use what Facebook has, such as friends who notice first, Groups and content signals, designed with clinical partners and safe-messaging practice - Privacy, consent, false positives and escalation, including private messages and minors - Outcome and harm metrics, and a testing approach that never withholds help ### Follow-up Questions - A post is flagged as signaling imminent risk in a country where Meta has no crisis-line partner. What should happen next? - How would you test whether a new outreach message works when you cannot hold back help from a control group? - Should Facebook limit feed time for people who seem vulnerable, even if engagement drops? How would you decide?