Russia has integrated a machine-learning system for filtering internet traffic into its existing network-control infrastructure, Meduza reports, citing Forbes Russia. The reported 2.27 billion-ruble system is a concrete expansion of state internet filtering, not just another plan to build one.
Watch Desk analysis
What happened
Meduza says the system runs on a Russian-made model and in-house infrastructure, and supplements TSPU, the existing system used to filter traffic on the Russian internet. It attributes the details to a passport for Roskomnadzor’s digital transformation programme covering 2026 and the 2027–2028 planning period.
The reported cost is 2.27 billion rubles. Meduza says the document gives no further technical details. The report also recalls earlier plans for AI-based filtering, but says Roskomnadzor had then been expected to use Astra AI, a model developed by Astra Group. The account does not establish whether that model was ultimately used.
Why it matters
This is a reported deployment of machine learning within a state internet-filtering system, rather than a proposal still waiting to be built. That makes the practical question how the system changes the reach or operation of existing controls. The available account does not describe its targets, accuracy or day-to-day use, so those consequences remain unclear.
The reported 2.27 billion-ruble price also puts a figure on the project. Meduza says Russia’s federal budget earmarks 60 billion rubles for traffic filtering over the next three years. Those figures suggest filtering is a substantial programme, though they do not tell us how much this particular system contributes to its results.
Our read
The important news is the reported integration of machine learning into existing filtering infrastructure, not the mere appearance of “AI” in a programme document. If the system is operating as described, the distinction between a filtering rule and a model-driven decision could matter for what gets blocked and how consistently. The missing detail is not a footnote: without it, claims about the system’s reach or effectiveness would be guesswork.
What to watch
- Whether more detail emerges about what the system is designed to identify or block.
- Whether Roskomnadzor or Forbes Russia publishes technical or operational information.
- How the new system is used alongside TSPU and other filtering tools.
Discussion spark: Should governments be required to disclose what machine-learning systems they use to filter internet traffic, even when they classify the technical details as sensitive?
Sources and evidence
- Russia rolls out AI-powered Internet traffic filtering system at a cost of 2.3 billion rubles – Meduza (13 November 2026, 09:34 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.