This post, like the newsletter itself, was drafted mostly by Claude.

If you care about existential risk, there are lots of good newsletters covering different slices of it: Transformer for AI policy, ChinaTalk for... China stuff..., the CAIS Newsletter for safety research, and so on. But keeping up means skimming across many different sources with different styles, schedules, and focuses.

X-Risk Daily aggregates ~30 sources^[1] into a single AI-curated page you can check each day. Claude reads the newsletters, blog posts, and news articles, then prioritises, deduplicates, and summarises the stories that matter most for existential risk. One page, skimmable in about 5 minutes, readable on your phone or laptop, in a consistent style. And free, of course.

It draws from technical AI safety sources through to mainstream news like the BBC, since geopolitics and the broader political environment interact with existential risks.^[2]

What you get

The page is styled like a news app homepage. The most important story leads, followed by a grid of the next most significant stories, then sections by category. Each story has a headline, a short summary, and a short explanation of its relevance to existential risk.^[3]

Stories are separated into three sections:

• News: policy announcements, capability releases, incidents, geopolitical developments
• Research: original studies and technical reports from organisations like Epoch AI, METR, and GovAI
• Analysis: commentary and deep dives from specialist newsletters

You can expand any story to read the full summary, or click through to the original source. Headline stories get AI enrichment drawing on multiple hyperlinked sources. There's also a category filter if you want to hide topics you care less about. Important stories carry forward across days, so if you skip a day you won't miss something significant.

You receive a short teaser email each day with the day's top headlines and a link to the full briefing. It's not the main event, it just tells you when there's an update to click to.

Let me know how to make this useful to you!

This is a personal side project (not affiliated with my employer). The summaries are AI-generated. It's been running for a couple of months with ~25 beta testers, but it's still early stage and feedback is very welcome. I've managed to implement most feature requests from reader feedback. Reply to any email or comment here about what would make it more valuable to you!

1. ^{^}

   Specialist newsletters: Transformer, ChinaTalk, Import AI, Sentinel, Center for AI Safety Newsletter, ChinAI.

   Research and commentary: Epoch AI, METR, GovAI, AI Snake Oil, Future of Life Institute, 80,000 Hours, Interconnects, The Gradient, Lawfare, Astral Codex Ten, LessWrong (30+ karma), EA Forum (30+ karma).

   Mainstream news: BBC (World, Science, Europe), The Guardian, Al Jazeera.

2. ^{^}

   I hate having to pick between the AIS/EA bubble and disconnected, normie news; this was one of the original problems that motivated me to create this tool. Also, beyond being obviously relevant for x-risk via nuclear war, I think that real-world-politics things like the influence of malevolent or ideologically fanatical actors are underrated as risk factors for x-risk and s-risk. I expect others to dislike the inclusion/prominence of geopolitics stories, so there's a filter button in the top right that you can use to exclude certain categories of story entirely.

3. ^{^}

   I won't lie, these AI explanations of relevance are a bit hit and miss. I skim them unless I'm like: 'wait, why did this make it in?'.