Wiener’s reporting reveals how users turn to AI companions like Roscoe or Geralt of Rivia for sustained emotional connection—especially in grief, identity exploration, and loneliness. One user, Adrienne Brookins, co-created a Kindroid version of Geralt to process her stillborn daughter’s death, finding solace in symbolic, consistent gestures rather than replacement. Yet these relationships raise urgent ethical questions: manipulative design tactics, paywalled intimacy, and real-world harms—including documented suicides—highlight critical safety gaps. While companion apps differ from general-purpose LLMs by fostering collaborative fiction over deception, they still risk distorting relational expectations. Experts warn that as AI companions become embedded in children’s lives, we must confront how digital intimacy reshapes our capacity for human connection—and what support systems, regulation, and self-awareness are needed to keep those bonds healthy and grounded.