Autonomous defence for decentralized networks

Autonomous Defence for
Decentralized Networks.

Adversarial AI is attacking validator infrastructure. We defend it with adversarial AI - trained on the network itself, acting at machine speed, only when authority has been earned.

Read the research

Open-source nodes. Proprietary intelligence. No black boxes.

[01]·The threat is here

The wave is here, not coming.

AI agents now discover and chain zero-day exploits autonomously, at $50 a vulnerability. Validator infrastructure - trillions in staked capital, hundreds of independent operators, no central SOC - is uniquely exposed.

April 2026 · Production exploit

Litecoin MWEB · 13-block reorg

A DoS-plus-consensus exploit on Litecoin's MWEB caused a 13-block reorganisation with roughly $600K in cross-chain exposure. The patch had been available for 37 days. Not a zero-day - a known-day that defenders couldn't close in time. CoinDesk →

December 2025 · DDoS wave

Solana absorbed 6 Tbps. Sui degraded.

Two major decentralized networks were hit with sustained DDoS campaigns in the same week. Solana - hardened by years of public outages - absorbed a 6 Tbps attack with no downtime. Sui - younger, less battle-tested - saw block production delays and degraded performance. Same wave. Different outcomes. The difference was the defensive substrate underneath. Most chains don't have one. CoinDesk →

[03]·The substrate

Normalising the abnormal.

There is no MITRE ATT&CK for validator infrastructure. No CVE pipeline. No shared format for what an attack on a decentralized network actually looks like. The industry has no substrate.

We built one. An open multi-modal bundle format, an open ten-family taxonomy, and a proprietary corpus of 1,092 contract-validated attack bundles across 19 primitives. Open the format. Keep the data. The same playbook that built every successful intelligence ecosystem.

Read the research →
[04]·Earned autonomy

Trust by evidence, not assertion.

Most autonomous defence asks for blind trust. We don't. IBSR runs in shadow mode on your traffic, producing counterfactual evidence of what it would have done. You decide when the evidence is sufficient.

Authority is granted in stages, scoped to specific abuse classes, and revocable at any time. The framework is published, peer-reviewable, and open.

Read the framework →

Deploy in shadow mode.

No enforcement. No risk. Counterfactual evidence on your traffic, your network, your terms.