Trail of Bits

Critical RCE chain across Vault and OpenBao patched for the latter, great work from the team operationalizing and remediating. Thanks to Trail of Bits for the collaboration and Alexander Scheel for PoC, coordinating and delivering fixes, and detail of IBM's response to our efforts to disclose Vault vulns in the post.

Our Avalanche experience includes security reviews of AvalancheGo, libevm, and Subnet EVM for Ava Labs. We're bringing that work to builders through the new Avalanche Audit Marketplace, where we're one of the vetted firms. We're also the team behind Slither, Echidna, and Medusa, and we've published 391 blockchain security reviews that anyone can read. Request a quote: https://lnkd.in/djwVUXZd

OpenBao v2.6.3 is out! 🔐 Our latest 2.6 patch release is focused on security and bug fixes. Make sure to patch your instances accordingly! This update also includes fixes identified through the Patch the Planet initiative in partnership with Trail of Bits and OpenAI. Big thanks to all reporters, contributors, and the entire community! 🔗 Read the full changelog here: https://lnkd.in/e_CZjAUQ #OpenBao #SecretsManagement #OpenSSF #OpenSource #Security

Threshold signatures secure billions of dollars in crypto, and teams increasingly run them inside Trusted Execution Environments (TEEs) for extra protection. But MPC and TEEs make different bets on trust: MPC spreads trust across independent parties, while TEEs concentrate it in the hardware manufacturer and its attestation infrastructure. In one pattern from our audits, a malicious host rolls back the filesystem after a signer deletes a used pre-signature. The signer reuses that value and leaks their private key share. We wrote up what TEE attestation can and can't fix in MPC deployments, the pitfalls we see most often in audits, and how to combine the two correctly. https://lnkd.in/eipeC8C9

Our OpenClaw Foundation security assessment produced 27 advisories, 3 hardening PRs, and 1 architectural submission. Read the report: https://lnkd.in/gBu_Husw

We're the only red team for Advanced Research + Invention Agency (ARIA)'s Safeguarded AI programme, led by Nora Ammann. Seven of the programme's eight teams are building software and hardware components with machine-checked proofs. In each development cycle, we'll test whether those guarantees survive realistic attacks. A proof guarantees exactly what its specification says, under the assumptions it makes. So we go past the implementation into everything the proof rests on: the models, specifications, assumptions, and deployment choices behind each team's assurance case.

We're hiring a Senior Developer Relations Engineer, and we want a builder. The person in this role ships software people use, contributes alongside our engineers and open-source maintainers, and uses AI tooling in daily practice. Their public work, whether research, talks, CVEs, or maintained projects, holds up with practitioners who check every claim. The job is to help our research and tools reach the people who need them: attract outside contributors, support maintainers, and bring what the community tells us back into the firm. If that describes you, or someone you'd vouch for, apply here: https://lnkd.in/eazF7FBS

Bee is a wearable that listens to your day. It's designed so that no one else, including Amazon, can read what it hears. We tested that design with a focus on the hardest attacker to stop: an insider with operator access. We found several ways an Amazon operator could run unapproved code on the attested servers that process user data. Amazon closed each one, fixed all 7 high-severity issues, and published the full report: https://lnkd.in/g4scAuSm

For the last six months I've been working on an amazing NVIDIA team that has been contributing security fixes to OpenClaw. Last week Trail of Bits completed an assessment on the project where they found 0 critical issues, 2 highs and a handful of lows and mediums. For a project that moves at the velocity of OpenClaw (the 2026.8.1 release contained 16k Pull Requests) this is an incredible success. Our team has reviewed over 1000 security incidents on OpenClaw in the last 6 months and contributed hundreds of fixes. Sometimes these fixes break functionality in unexpected ways and we are sorry for that, this is the delicate balance of security remediation work. We will continue to work to avoid regressions like these in the future. For a project that started the year in infamy due to the challenges of securing personal agents things have come a long way. I'm really excited to continue contributing to OpenClaw as it matures into production software. Links in the comments 👇

"Could've cost millions." That's how Trail of Bits refers to a bug they found in the Miden VM core library while testing our systems from the ground up. The key word is could've, because they found it, and fixed it. Miden VM runs on its own assembly language, MASM. When Trail of Bits first looked at MASM, it had almost no developer tooling: no editor support, no linter, no static analyzer. So they spent six months building it all from scratch, before the review even started. AI helped. ToB's agents, both Claude and Codex, built an editor plugin, a decompiler, a static analysis engine, and a formal model of the VM in Lean, under human supervision. That’s 100+ commits of tooling that didn't exist, enabling humans to go deeper than any manual review, Trail of Bits says. The results: 🔸A high-severity signature-forgery bug, caught and fixed pre-launch 🔸95 machine-checked correctness proofs covering the core library's binary arithmetic 🔸A static analysis engine we've adopted permanently -- every future change gets the same scrutiny Trail of Bits looked at where security reviews are headed in the age of AI. What can agents build and how do you manage them? And, why the economics of deep audits just changed. https://lnkd.in/eD75A3mv