1.5K ·GitHub
What is this?
AI Safety HOT Hub is a daily-updated feed of curated AI safety news and research papers (attacks/jailbreaks, defenses, alignment evals, incidents, multi-agent risks, governance), with Markdown/BibTeX/JSON exports and a public MCP endpoint so an AI agent can query news, papers, events and daily briefings directly.
AI Safety HOT Hub is a daily-updated feed of curated AI safety news and research papers (attacks/jailbreaks, defenses, alignment evals, incidents, multi-agent risks, governance), with Markdown/BibTeX/JSON exports and a public MCP endpoint so an AI agent can query news, papers, events and daily briefings directly.Is this a tool I install, or a hosted service I connect to?It is a free public hosted MCP endpoint plus a companion website; there is nothing to install locally, you just point your agent client at the remote MCP address.
It is a free public hosted MCP endpoint plus a companion website; there is nothing to install locally, you just point your agent client at the remote MCP address.How do I connect my agent to it?Run the provided connect command for Codex or Claude Code, or for other clients add an MCP server named 'aisafetyhot' using the Streamable HTTP address, then restart the client session.
Run the provided connect command for Codex or Claude Code, or for other clients add an MCP server named 'aisafetyhot' using the Streamable HTTP address, then restart the client session.Do I need an API key or account to use the MCP?No, the MCP is public and requires no login or API key.
No, the MCP is public and requires no login or API key.Do the daily report and the paper archive cover exactly the same papers?No - papers are archived by Beijing-time calendar day, while the daily digest covers the prior day 08:00 to current day 08:00 window, so a paper can show up in an adjacent day's report versus its archive date.
No - papers are archived by Beijing-time calendar day, while the daily digest covers the prior day 08:00 to current day 08:00 window, so a paper can show up in an adjacent day's report versus its archive date.What topics does the feed cover?It is organized into six categories: attacks and jailbreaks, defenses and guardrails, alignment and safety evaluation, AI incidents, multi-agent insecurity, and governance and policy.
It is organized into six categories: attacks and jailbreaks, defenses and guardrails, alignment and safety evaluation, AI incidents, multi-agent insecurity, and governance and policy.How often is new content added?Daily reports are published every day at 08:00 Beijing time, and the Hub checks for updates every 15 minutes.
Daily reports are published every day at 08:00 Beijing time, and the Hub checks for updates every 15 minutes.Where does the underlying framework come from?AI Safety HOT is built on the open-source AIHOT framework, with credit given to its original author.
AI Safety HOT is built on the open-source AIHOT framework, with credit given to its original author.What can I do with the downloadable paper lists?Each day's papers since 2026-09-24 are archived as Markdown (titles, summaries, AI-generated readouts), BibTeX (for Zotero/EndNote import), and JSON (for agents or custom scripts).
Each day's papers since 2026-09-24 are archived as Markdown (titles, summaries, AI-generated readouts), BibTeX (for Zotero/EndNote import), and JSON (for agents or custom scripts).What are similar projects?AI-Papers-of-the-Week, chubbyskills, awesome-agent-skills-security, Horizon, Awesome-GitHub-Repo, weekly, awesome-mcp-servers, Agent-Learning-Hub
AI-Papers-of-the-Week, chubbyskills, awesome-agent-skills-security, Horizon, Awesome-GitHub-Repo, weekly, awesome-mcp-servers, Agent-Learning-HubWhich topics is it in?newsletter, governance, tex, news, rss, arxiv, papers, dataset, model-context-protocol, knowledge-base, json, china
newsletter, governance, tex, news, rss, arxiv, papers, dataset, model-context-protocol, knowledge-base, json, china