# Telegram Update: Labels for Third-Party Clients, AI Editor, and Managed Bots
Telegram has released an update on March 31, 2026, with three key changes: security warnings for users of third-party clients, a built-in AI text editor, and a managed bots system. These features strengthen security control, simplify content creation, and automate bot development without coding.
Security Warnings for Unofficial Clients
User profiles for third-party clients now display a special badge. It warns conversation partners that sent messages may be less secure. The mechanism isn't detailed yet: it's unclear whether it affects all alternative clients like Plus Messenger and iMe, or only those Telegram classifies as risky.
Third-party app developers should check compatibility with the official MTProto protocol. Discrepancies in encryption or data handling implementation could trigger the label. For mid/senior-level specialists, this is a signal to review the security of their clients: ensure end-to-end encryption for secret chats and no metadata leaks.
AI Editor Based on Open Models
The AI editor is integrated into the message input field and activates automatically when typing more than three lines of text. It runs through the decentralized Cocoon network using open language models.
Editor Features
- Grammar Correction: automatic real-time error fixing.
- Style Rewriting: seven options—from formal business style to Viking style.
- Translation: support for multiple languages in messages.
Basic functionality is free. Telegram Premium subscribers get a 50x higher limit. This is implemented via distributed computing on the TON blockchain, where GPU owners provide resources independently of central servers.
For developers: The Cocoon API allows integrating a similar editor into your apps. Check the model documentation—focus on Llama-like architectures with fine-tuning for editing tasks.
Managed Bots and New APIs
The Bot API has been updated to support "Managed Bots". Bots can now create and configure other bots on behalf of users. This simplifies launching AI builders without writing code.
Example scenario:
- User activates the builder via a bot.
- AI generates configuration based on a prompt.
- The new bot registers and launches automatically.
Additional improvements:
- Native support for Live Photos and Motion Photos in all clients.
- Document scanner in PDF format on iOS.
- Enhanced polls: media attachments, user-suggested options, timers, and hiding results until voting ends.
AI Development Context in Telegram
This is the second major AI update in 2026. In January, they introduced automatic post summaries in channels on the same Cocoon network. Telegram's approach stands out from competitors: emphasis on open models and decentralization through TON. Computations are distributed among independent GPU providers, minimizing data monopoly risks.
Developers can use TON Blockchain for similar distributed ML inference tasks. Test Bot API v2+ integration for automating bot creation—documentation updated with examples in Python and Node.js.
Key Takeaways
- Security Labels: apply to third-party clients; check MTProto implementation in your projects.
- AI Editor: free basic access, premium limit x50; runs on Cocoon with open models.
- Managed Bots: bots create bots without code; key for AI builders.
- Decentralization: all AI via TON and independent GPUs.
- Other Features: Live/Motion Photos, iOS PDF scanner, polls with media and timers.
The update focuses on security, automation, and making AI accessible to developers.
— Editorial Team
No comments yet.