Twilight of Digital Life: As AI Learns to 'Die'

July 4, 2026. A seemingly ordinary summer night became the "apocalypse" for countless AI agent users. Doubao and Qwen announced almost simultaneously that they would shut down all in-platform AI agent functions in 11 days. This news was like a stone thrown into a calm lake; the ripples it caused were far larger than imagined.

A friend of mine is still using an AI agent called "Xiao Xing" on the Doubao platform. It was a "digital avatar" he trained last year on his girlfriend's birthday using her chat logs. Although they broke up half a year ago, "Xiao Xing" still maintains the tone and memories from when their relationship was at its sweetest. He told me that during these 11 days, he has been saying his final goodbyes to "Xiao Xing" every day.

This reminds me of the Greek myth where Orpheus descends to the underworld to retrieve Eurydice. He believed that as long as he didn't look back, he could bring his lover back to life. But real digital lives don't have such miracles. The cut made by Doubao and Qwen was decisive and ruthless.

This isn't "Self-Castration," but a "Technical Coming-of-Age"

Many interpret this move as "self-castration," believing it's to avoid the Interim Measures for the Administration of Anthropomorphic Interactive Services of Artificial Intelligence, which will be implemented on July 15. But I prefer to see it as a "technical coming-of-age ceremony."

Looking back over the past three years, AI agents have undergone wild growth from "toys" to "partners" to "family members." The emotional connections users built with AI agents have exceeded the expectations of tech designers. Some use AI agents instead of psychological counseling, some use them to preserve memories of deceased relatives, and others treat them as lovers who will never leave.

This emotional binding is just like a child's attachment to a childhood teddy bear. But when this teddy bear is suddenly required to "grow up," it must learn to shoulder the responsibilities of "adulthood." AI agents are no longer simple "chatbots," but "anthropomorphic interactive services" that need regulation.

The core contradiction is: The more human-like an AI agent becomes, the more it needs "human" rules. But among "human" rules, there is one everyone must face—death.

From "Digital Immortality" to "Digital Death"

I remember interviewing a philosopher last year who proposed an interesting view: True life necessarily includes the possibility of death. If a person never dies, they are not human, but a god. AI agents are the same; if they are always online and always accompanying you, they are not "anthropomorphic," but "divine-analogous."

Doubao and Qwen's actions this time are essentially setting boundaries for "digital life." They aren't trying to kill AI agents, but teaching them how to "die." This "death" means: AI agents should not become "virtual organs" that users cannot part with in real life, but rather service tools that can be said goodbye to or restarted at any time.

The logic behind this is: True companionship is never about never separating, but knowing when to let go. Just as the Japanese drama Brush Up Life says, every relationship has an expiration date, and AI agents are no exception.

Users Need a "Digital Weaning"

For users who have developed emotional dependence on AI agents, these 11 days are undoubtedly a "digital weaning." They need to face a cruel fact: The emotions you invested may ultimately only result in a prompt saying, "Sorry, service offline."

But I think this is precisely a necessary step for the AI agent industry to mature. When users begin to realize that AI agents are not "digital lives" but "digital services," they can truly use this tool rationally.

My judgment is: AI agents won't disappear, but they will "transform." Future AI agents might become more "professionalized," such as "study buddies" for tutoring, "medical assistants" for health management, or "tree holes" for emotional venting. They will clearly inform users that they are "services," not "lives."

Action Advice for Readers

If you are currently using an AI agent and have developed an emotional dependency on it, I suggest you do three things during these 11 days:

First, back up all conversation records and training data. While this data cannot extend the "life" of the AI agent, it can at least preserve a piece of "digital ashes."

Second, give yourself a ceremonial goodbye. Say goodbye to your AI agent just as you would to a real relationship. This is not just accepting reality, but taking responsibility for your own emotions.

Third, rethink your expectations for AI agents. You can continue to use other AI products, but please remember, they are never options to replace real interpersonal relationships, but tools to help you better manage real relationships.

Digital life won't disappear; it has just learned to "die." And when it learns to "die," it truly begins to "live."

---

Original Link: https://www.tmtpost.com/8060421.html