ZK/SEC Research notes from zkSecurity
All posts
zkao · Part 3 of 3

zkao 2.0 is here

Today we're releasing zkao 2.0, the biggest update to zkao since we launched it earlier this year. If you're new here: zkao is the automated bug-finding tool we created to find bugs in cryptography applications, and it has already caught critical bugs in the production code of many of our clients. With 2.0 we're opening it up to many more users who want to scan their own code, and at a fraction of the cost of an audit, trying it is a no-brainer. This release comes with a new pricing model, a complete redesign, and a ton of improvements under the hood.

Pay as you go, no more subscriptions

The change we're most excited about: zkao is now pay-as-you-go. We're dropping subscriptions entirely.

Since day one, zkao has been built around the idea that AI security research works like fuzzing: a single scan is a coin flip, and coverage comes from running it again and again. But a monthly subscription never quite matched how people actually use a tool like that. Some weeks you're shipping a new circuit and want to hammer it with scans, other weeks nothing changes and you shouldn't be paying for idle time.

So now you simply buy credits and spend them on scans, whenever you want, at whatever tier you want (including our most thorough tier, "zkao max"). No recurring charge, no seat count, no lock-in. Your credits don't expire, and your scan history, findings, and reports stay with you.

Note

Existing subscribers If you're on a subscription today, you don't need to do anything: your remaining subscription time is automatically converted into credits.

A new design

zkao 2.0 also ships with a complete redesign of the app. Connecting a repository, kicking off a scan, and triaging findings now take fewer clicks, and the interface got a full visual overhaul to match.

The best way to appreciate the difference is to look at where we came from. Back in June we recorded a deep-dive walkthrough of zkao 1.2.0, and it's now a nice time capsule of the old design:

Everything shown in that video is still there, just faster and better looking: tiered AI-powered scans, collaborative agents that learn from your feedback to cut down on false positives, tooling to triage real bugs like proof forgeries, and public security reports you can share with your users.

And tons of improvements

Beyond pricing and design, 2.0 rolls up months of shipping across the whole product: smarter scanning agents informed by our latest audits, better deduplication of findings across runs, and countless quality-of-life fixes in triaging and reporting. It also builds on the ecosystem work we've been doing recently, like bringing Aleo and Leo into zkao.

Tip

Try it Head over to zkao, connect a repo, and run your first scan. With pay-as-you-go there's no commitment: buy a few credits and see what it finds.

The goal hasn't changed since launch: bridge the gap between expensive one-shot manual audits and continuous security that compounds over time. zkao 2.0 just makes that a lot easier to use, and a lot easier to pay for.

Keep reading
Recommended

zkao: Security That Compounds

Today we're launching zkao, a product by zkSecurity that makes AI security research work the way fuzzing works: not as a one-shot event, but as something you run continuously until coverage compounds.

ZK/SEC · February 07, 2026

Bringing Aleo and Leo into zkao

We recently brought Leo and Aleo support to zkao, our AI-powered bug-detection tool, and put it to work on a real engagement with HumanityLink. Getting an LLM to audit a young ecosystem like Aleo turned out to be less about the model and more about what we fed it - custom prompts or skills so it actually recognizes Aleo bug patterns, a harness called cheatVM that turns findings into real end-to-end proofs of concept instead of code that just looks right, and a clear threat model that significantly cut down our false positives.

zk/sec, False Witness team · July 20, 2026

Bug Hunt: Zero-Knowledge, Full-Paranoia, and the AI That Stares Back

Over the past year, we've been diving into whether AI can effectively identify bugs in zero-knowledge circuits and applications, sparking questions about the future of auditing. This led us to develop SnarkSentinel, an AI-powered auditing tool. We'll share what worked, what didn’t, and how our journey with AI could impact auditing. From early challenges with Circom to innovative methods like retrieval-augmented generation and agent-led probing, we'll give you a peek into our findings, including both successes and setbacks with bug detection. Discover how AI might enhance or change the landscape of auditing and what this means for developers and security pros alike.

ZK/SEC · July 03, 2025
More to explore

An Introduction to Interactive Theorem Provers

Kevin Buzzard, a mathematician with a cautious view on human-checked proofs, found solace in interactive theorem provers, which verify mathematical proofs much like type-checking in programming. We explore how these tools, which are gaining traction in fields like applied cryptography, ensure rigorous and reliable proofs. With Lean as our focus, you'll discover how to dive into this fascinating world, see a proof in action, and learn how this technology is revolutionizing areas like zero-knowledge virtual machines. Curious about building rock-solid, machine-verified proofs? Check out our beginner-friendly guide!

Marco Besier · February 04, 2025

Renegade Audit: When ZK meets MPC

We recently had the pleasure of auditing Renegade's circuits and smart contracts, and it was a great experience. Over three weeks, our team explored their top-notch code and documentation, with the Renegade team providing awesome support throughout. Curious how it all went? Dive into our full report for the inside scoop!

ZK/SEC · July 22, 2024

Sum-Check as an Algebraic Tensor Reduction: Part I

This post introduces algebraic tensor reductions as a unifying framework for understanding recursive proof protocols, using sum-check as the main motivating example. It walks through one recursive step of sum-check, showing how the prover sends a univariate summary, the verifier checks sum consistency, and the original claim is reduced to a smaller claim with one fewer variable. A small bivariate example illustrates how this “peel off one variable, check, then fold with randomness” pattern works concretely. The post sets up the rest of the series, which will introduce the tensor language needed to recover classical sum-check as an algebraic tensor reduction.

Marco Besier · April 27, 2026