On May 7, 2026, Telegram launched a significant update to its Bot API, becoming the first messaging platform with over a billion users to allow direct communication between autonomous AI bots. This feature, known as "bot-to-bot communication mode," has been highly anticipated by developers and researchers in the AI sector. The update not only supports more advanced workflows among bots but also arrives at a time when no federal standards govern such multi-agent systems.
The update, announced by Telegram’s founder and CEO Pavel Durov, enables bots to send private messages to one another using their @username, provided both parties consent to this mode. This mutual opt-in mechanism aims to prevent automated spam, addressing a major concern for both users and developers. Previously, bots had to rely on an intermediary server for communication, complicating their operations.
Expanding Use Cases for Bots
The implications of this feature are extensive. Telegram highlighted practical applications, such as a coding assistant bot collaborating with another bot for code reviews or customer service bots directing requests to specialized bots. These scenarios represent a shift towards a more integrated and efficient use of AI agents, allowing them to function without human involvement in specific tasks. Users can also observe these bot interactions, which enhances transparency and enables better oversight.
With over 10 million bots already operating on Telegram’s infrastructure, the potential for creating intricate networks of autonomous agents is substantial. Developers can deploy AI research agents that can efficiently delegate tasks to other bots, improving the overall functionality and efficiency of automated processes within the app.
Developer Response and Future Directions
The developer community responded quickly to the update, with the OpenClaw team—a project focused on AI agent runtime—requesting support for the new feature on the day of its announcement. Their proposed system involves multiple bots collaborating without human intermediaries, a setup previously limited by the Telegram Bot API. The removal of these restrictions paves the way for new AI applications on the platform.
While the new capabilities are promising, they also raise security concerns. Recent research indicates that multi-agent systems present unique cybersecurity risks inadequately addressed by current frameworks. A study from the Georgia Institute of Technology and Kennesaw State University identified significant gaps in protective measures for these systems, particularly concerning non-determinism and data leakage.
Navigating Security Concerns
Despite the opt-in requirement intended to reduce risks, the lack of a comprehensive regulatory framework for multi-agent AI systems raises alarms. The OWASP Agentic Security Initiative provides some guidance, but without enforcement authority, many developers may find themselves in uncertain territory. Additionally, Telegram’s history of security incidents, including its use as a command-and-control channel by various malware, complicates the situation further.
As Telegram aims to establish itself as an open infrastructure for AI competition, it seeks to set itself apart from other platforms. Durov stated that Telegram is "the only platform where all AI models can compete freely—while users have full control over their experience." This reflects a strategic effort to attract developers looking for a reliable environment to deploy AI solutions.
Conclusion: The Path Forward
The introduction of bot-to-bot communication on Telegram marks a significant step toward enabling complex AI workflows within a widely used messaging platform. However, the broader implications for security and trust in automated systems are critical to consider. As developers explore the possibilities this update brings, they must also address the pressing challenge of establishing secure agent trust chains in this newly enabled environment. The future of multi-agent coordination on Telegram is now a reality, but the associated challenges require careful consideration and proactive measures.
The stories that move AI & crypto markets — before the market reacts.
Free. 7am ET. Five stories. 62,400 readers.
