About

AI Edge Briefing is a daily, fact-first briefing on frontier AI — the advances, the research, and how the technology is being used, for good and for harm. Cyber operations and influence campaigns, military and defence, health and science, policy and the courts, chips and compute. Every headline links to its source. Nothing is written without one.

Why this exists

I'm Mike Shoss. I lead Epilogue, an AI consulting and product studio in Toronto. Part of that job is knowing, every morning, where this technology actually is — so I've long started the day the same way: the lab announcements, the new papers, the threat-intelligence reports, what's landing in defence, health and policy.

It's a job requirement, but it's also a fascination. This is the most powerful technology we've ever had, and every day it's being used to do remarkable good and real harm — often in the same report. What pushed me to publish was Anthropic's September 2026 threat-intelligence report and the realisation that most people who should know what's in documents like that never will. Not everyone can keep up, or wants to, or has the time. I can, and I felt it was on me to share it where I could.

So this is the briefing I was already making for myself, made public.

How an edition is made

Every morning:

• The sweep. About 170 sources — labs and their research blogs, arXiv, security vendors' own threat reports, government and regulator sites, court dockets, defence and health press, and the trade and mainstream outlets that cover them. The full list is public.
• The rules. Primary sources first. Every number quoted exactly as the source wrote it, with its baseline. Every caveat the source raises is stated — company claims, single-source stories, preprints — and flagged as such. If a claim can't be traced to a document that can be opened, it's left out. Opinion without new facts, marketing without numbers, and rumour don't make the cut. The editorial rules are public too.
• The check. Every link is tested live before publishing; a dead link fails the build.

Mondays add a week in review: the threads that mattered, the week's key figures, and what's on the calendar.

The role of AI — read this part

The research, the writing and the voices are produced by AI. Claude does the morning sweep, verifies items against their primary sources and writes each edition; OpenAI's speech models voice the podcast. No person reviews an edition before it goes out.

What makes that trustworthy isn't the model — it's the checks around it, and the fact that you can see everything:

• Every edition publishes its complete build log — every page fetched, every search run, every check passed — recorded automatically by the tooling, not written by the AI. Open any edition and click how this edition was made (here's the latest).
• The podcast script is mechanically locked to the edition. Before an episode can be made, a validator rejects any number that doesn't appear in the source item, any missing caveat, any unnamed source, and any speculative or hyped language. If a day's script can't pass, that day is narrated straight from the edition text instead.
• Every claim links to its source, so you never have to take the briefing's word for it.

It will still get things wrong sometimes. When it does, tell me: [email protected]. Corrections are made in the open.

The podcast

The AI Edge, presented by Epilogue, is each day's edition as a 10–15 minute conversation between two hosts, Maya and Alex. They're AI voices, and they say so at the top of every episode. Each episode has a transcript showing exactly which item every part of the conversation came from. Listen on Spotify or add the RSS feed to any podcast app.

Each episode's cover is generated from that day's news: every section has a colour — security is red, research violet, military orange, and so on — and the cover mixes them in proportion to how much of the day fell in each.

Trends

A topic is trending when it keeps appearing across editions. The Trends page shows what's recurring right now, and every topic has a page collecting every item ever filed under it — the easiest way to follow one storyline over weeks.

Who's behind it

Epilogue is an AI consulting and product studio based in Toronto, focused on turning complex business problems into practical, high-impact AI products — built end to end.

Mike Shoss is Epilogue's founder and principal. He ships AI products in industries where being wrong is expensive, advises and mentors AI startups at The Forge and DMZ, and is an angel investor in early-stage Canadian AI. He lives in Milton, Ontario. LinkedIn

Open

The editions, the source list, and all the code are on GitHub. The editions are published under CC BY 4.0 — reuse them with a link back. The code is MIT. Full terms.