AI Bots Can't Use WhatsApp Anymore. So Who Are They Going to Talk To? (www.process-one.net)

Meta announced that, starting January 15, 2026, WhatsApp will ban all third‑party general‑purpose AI chatbots (e.g., ChatGPT, Perplexity, Poke), leaving only Meta’s own AI on the platform. Meta frames the move as a response to operational burdens, but it effectively centralizes control of the AI layer inside its messaging walled garden. By contrast, Discord—without building its own AI—expanded server capacity to 25 million as third‑party tools like Midjourney (now ~20M members in a single server) use Discord as their interface, showing a platform-first strategy that benefits from hosting independent AI agents rather than owning them. The decision highlights a critical inflection point for AI/ML: platform control vs. open interoperability. If messaging platforms lock out third‑party agents, we risk creating incompatible “agent ecosystems” that mirror past app‑store monopolies, complicating use cases that require cross‑domain agent collaboration (procurement agents negotiating with supplier logistics agents, or federated healthcare AIs across hospital systems). Standards work is already underway—the IETF and the A2A initiative are drafting agent‑to‑agent protocols—but the window to design interoperable defaults is closing. The technical and policy choices made now will determine whether agents remain composable across platforms or become siloed by whichever vendors move fastest.