Lighthouse

Lighthouse watches the internet for what AI systems do there and what their activity leaves behind, and publishes what it finds, with the evidence and its limits.

Short form

Who watches the internet?

Released

If you wanted to know whether something was going wrong with the internet itself, whose records could you read without paying? In November 2026 we read the public pages of every instrument we could find, sorted across ten layers from cables to traffic, and kept 48.

By what we found, every layer has at least one. The gaps sit inside the layers: some come from where an instrument stands, some from what it counts, and some from who may read it. The two big public routing archives listen from inside about 1.3 per cent of some 74,000 networks, by a 2024 preprint's count. The one instrument we read that claims to cover the IPv6 address space broadly says in its own paper that it doesn't scan it comprehensively. Traffic is seen one network, link or sensor fleet at a time, or in aggregate. The data behind one daily set of exploit-prediction scores "cannot be released", its makers say. And 11 of the 48 build on another of them, so there are fewer separate vantage points than instruments.

The main limit: we read documentation, not data. This is a first sketch from what the instruments say about themselves, not a census, and it measures nothing of the internet itself.

The full piece

From articles/short/who-watches-the-internet.md in the repository. Published 7 November 2026.