Awesome AI Security Telegram [](https://awesome.re)
December 11, 2025 · View on GitHub
Curated list of Telegram channels and chats on AI Security, MLSecOps, LLM Security and AGI Security.
Initial seed: the AI security folder
https://t.me/addlist/l9ZMw7SOW9hjYzUy
Language legend: RU = mostly Russian content, EN = mostly English or mixed.
Contents
- Tier 1: Core AI Security Channels (T1)
- Tier 2: Extended Ecosystem (T2)
- Community Chats
- How to Contribute
- Badges & Telegram Widgets
Tier 1: Core AI Security Channels (T1)
These are high-signal, author-driven channels that consistently publish deep content on AI security, LLM security, MLSecOps and related topics.
| Channel | Lang | Focus |
|---|---|---|
PWN AI  | RU | Practical AI Security and MLSecOps: how to apply AI in security engineering, how AI breaks, and how to defend it. Strong focus on LLM security, agents, guardrails, real-world threats. |
| Борис_ь с ml | RU | Machine Learning + Information Security: author blends ML, data science and cyber/AI security; long-form analyses, conference recaps, and practical viewpoints from a vendor-side security practitioner and methodologist. |
| Евгений Кокуйкин — Raft | RU | Building Raft AI and GPT-based applications. A lot of hands-on posts on building products with LLMs, including trust & safety, reliability and security concerns. |
| LLM Security | RU | Focused on LLM security: jailbreaks, prompt injection, adversarial attacks, benchmarks, defenses for LLM-based apps and agents. Good for monitoring the global LLM security landscape. |
| AISecHub | EN | Global AI security hub: curated research, articles, reports and tools across adversarial ML, LLM security and AI governance. Strong link to the broader AISecHub ecosystem. |
| AI Security Lab | RU | “We hack AI so others cannot.” Laboratory by Raft x ITMO University x AI Talent Hub. Mix of course materials, research and hands-on case studies on breaking and defending AI systems. |
Tier 2: Extended Ecosystem (T2)
Supportive channels that complement the T1 set: feeds, digests, tooling-oriented and topic-specific streams.
| Channel | Lang | Focus |
|---|---|---|
| ML&Sec Feed | RU/EN | Aggregated feed for ML & security: news, tools, research links across ML security, LLM safety, privacy and classic infosec. Good “signal booster” for the ecosystem. |
| AISec [x_feed] | RU/EN | Digest of AI security content from X, blogs and papers, without paywalls. Built for executives, MLSecOps and AI security experts who need material for decks and reports. |
| AI SecOps | RU | AI Security Operations: materials, links and events around operating AI systems securely in production (monitoring, incident response, SIEM/SOC integrations, etc.). |
| OK ML | RU | ML/DS/AI channel with explicit focus on repositories, tools and vulnerabilities in ML systems. Bridges classic ML engineering and ML security topics. |
| AI Attacks | EN | Stream of AI attack examples and threat intelligence: adversarial ML, LLM jailbreak demos, malicious AI use-cases, attack tooling. |
| Poxek AI | RU | AI-focused spin-off in the “Похек” universe. Currently low-traffic / experimental, expect content at the intersection of pentesting, Poxek community and AI/LLM security. |
| AISecurilka | RU | Niche AI security channel. Emerging space for short notes, links and experiments around AI/LLM security (description to be refined as it grows). |
| AGI Security | EN | “Artificial General Intelligence Security” — discussions and links around long-term AGI security, safety and existential-risk-flavoured topics. |
| paranoAISecure | RU/EN | Experimental channel on “healthy paranoia” for AI systems: threat modeling, secure design patterns, operational practices for AI security (early-stage, description to be clarified). |
Community Chats
Interactive spaces for Q&A, red teaming, experiments and community work.
| Chat | Lang | Focus |
|---|---|---|
LLAMATOR — AI Red Team Community  | EN/RU | Community chat for the LLAMATOR framework: red teaming and testing LLM vulnerabilities (prompt injection, jailbreaks, system-prompt leakage, harmful behavior benchmarking, etc.), supported by ITMO University and AI Talent Hub. |
| Безопасность ИИ и ML | RU | Open chat dedicated to security of artificial intelligence and machine learning; informal place to discuss attacks, defenses, tools and papers in Russian. |
How to Contribute
This repository is intended to become a living map of AI Security Telegram.
-
Suggest new channels/chats
- Open an issue or PR.
- Provide:
@handle, language, 1–2 sentence description, and why it is relevant to AI security / MLSecOps / LLM security.
-
Propose re-tiering
- T1 is for consistently high-signal, author-driven channels with strong AI security focus.
- T2 is for feeds, adjacent topics, niche or emerging channels.
- In your PR, explain why a channel should move tiers (signal, posting frequency, focus shift, etc.).
-
Keep it on-topic
- Focus on: AI security, MLSecOps, LLM/agent security, adversarial ML, AGI security and governance with a clear security angle.
- General ML, generic infosec or news-only feeds without AI security focus should stay out of scope.