About Signal.

What we do.

Signal publishes long-form writing on artificial intelligence. We cover the technical foundations — the mathematics, the architectures, the failure modes — alongside the cultural and political consequences of building machines that think.

We are not a news site. We do not chase announcements. We prefer to spend three months on a piece that explains one idea well, rather than three days on a piece that explains ten ideas badly.

What we don't do.

We don't publish hype. We don't publish doom. We don't run sponsored content, we don't do "5 trends to watch" listicles, and we will not pretend that any current system is sentient. If a piece sounds like a press release, we don't write it.

The job of a journal is to slow the conversation down, not to speed it up. — EDITORIAL CHARTER · 2026

How we work.

Every piece goes through three passes: a technical review by someone who has actually built the thing being described, a structural review by an editor who has not, and a final read for clarity and tone. If a piece cannot survive all three, it does not run.

We accept outside submissions on a rolling basis. We pay for published work. We commission roughly half of what we publish and accept the rest from writers we have come to trust.

Who we are.

Signal is edited by a small team based in Berlin, Toronto, and Singapore. We are writers, researchers, and engineers — most of us with day jobs in machine learning or its adjacent disciplines.

• Dr. Maya Chen — Deep Learning, Optimization. Toronto.
• Liam Park — Computer Vision, Robotics. Berlin.
• Sarah Okonkwo — AI Ethics, Governance. London.
• Editorial Team — Generative AI, NLP, Reinforcement. Singapore.

How to reach us.

For pitches: pitches@signal.journal. For corrections: corrections@signal.journal. For everything else, find us on the usual social networks — we post the work, not the discourse.

---

Last updated · August 16, 2026