Aether AI

We appreciate the hard work & effort from the team at OpenClaw for triaging and mitigating these findings and we're happy to see all of these have now been resolved.

We pointed Aether AI at OpenClaw and ended up with 16 confirmed CVE's. Credential theft, container escapes, shell injection, auth bypasses across multiple integrations. Full attack chains, executed autonomously, no human direction required. And it only took hours, not weeks.

@JustWantToQ1 @callmeveizir @theonejvo Currently our onboarding is processing verified organisations as preferential. However we'll be opening up for researcher access concurrently so any waitlist applications will be processed regardless. As to be expected, all applications manually/strictly vetted regardless.

@tryaether_ai @callmeveizir @theonejvo Do I need to be apart of a corp or can I just use a solid referral?

Anyone can open up Claude or ChatGPT and ask them to find security issues. They will try. Find a few, then they will forget. Then they will stop quoting safety concerns. @tryaether_ai is the purpose-built attack AI that runs autonomously for 114 hours, with full memory of every finding, every dead end, and every pattern from every prior pentest & vulnerability in your attack surface. Building that required solving problems that chat interfaces were never designed for.

@callmeveizir @theonejvo We will be considering researcher access, so your waitlist submission won't be ignored. It just may not be part of the first release group. All waitlist participants will be contacted so don't stress.

@theonejvo @tryaether_ai aah i have filled their waiting list but nvm

2 targets. 114 hours. 71 vulnerabilities. 1 CTO calling his pentest partner asking how they missed them all.

After spending the last 15+ years hacking some of the worlds most "secure" targets. From Governments & battlefield technology, to banks and even casino's. We're soon releasing the worlds most dangerous attack AI - to help you fight back and stay far ahead of your adversaries.