AI Sequrity Company

We did a write up about how one of our sandboxes running openclaw got recently compromised running absolutely benign evals — all because of a few ticks! substack.com/home/post/p-18…

Folks, we are hiring a few systems researchers/engineers (full-time, part-time or internships) with the following requirements: * Experience in systems research * Familiarity with inference stacks such as vLLM, SGL, or TensorRT * Python, CUDA, and Rust experience is a strong bonus

AI agents face a critical security threat: Indirect Prompt Injection (IPI) attacks hijack agent behavior to steal credentials and cause financial loss. We built the first system-level defense for Computer-Use Agents that actually maintains utility! That is, don't change the model, change the system around it to guarantee control flow integrity! arxiv.org/pdf/2601.09923

Live from EurIPS, the European NeurIPS gig. 🇪🇺 It was an honor to keynote at the LLM Security Workshop here in Copenhagen. I spoke about the current state of System-Level Defenses—moving beyond the idea that we can perfectly "align" a model into safety, and instead looking at how we secure the broader systems these models inhabit. EurIPS turned out to be good fun, bringing the European research community together to tackle these hard problems. Hope it sticks around! Thanks to everyone who joined the discussion!

Excited to be speaking at AI Defence EU Cyber Week today @EUCyberWeek ! I am discussing AI Security and how we can stay ahead of emerging threats. Come say hi if you are around

5 years ago, I wrote a paper with @wielandbr @aleks_madry and Nicholas Carlini that showed that most published defenses in adversarial ML (for adversarial examples at the time) failed against properly designed attacks. Has anything changed? Nope...

Thrilled to see our CaMeL, with @edoardo_debe, featured in the @stateofai by @nathanbenaich! While powerful, CaMeL is challenging to implement in practice. That's why we're excited to announce a new scheme from @aisequrity that provides strongest security guarantees that can be used with just five lines of code. Want to partner and try it out? Ping me at ilya@sequrity.ai

Our longer chat with @MLStreetTalk is live here: youtube.com/watch?v=aoX_pG…

RT @iliaishacked: Dear friends, we at @aisequrity are looking to hire a few first engineers to join us. Are you a Uk based student who want…

AI agents break every single assumption in the book. We need to talk about AI agents and security. And ... I don’t mean the usual "prompt injection bad". Traditional insider threat models assume rational actors with physical constraints.

My friends, today I am excited to announce AI Sequrity (@aisequrity). Our mission is to provide developers and enterprises a painless and stress-free deployment of AI that is secure by design. You heard this right. You can deploy your AI agents and get guaranteed security. The best part? To use our systems you need to change at most five lines of code. We are ready to do on-prem and already support extremely complex policy use-cases. We are already best-in-class and have a ton more features coming out soon. We are not selling yet another guardrail system, our tools are a lot more precise and enable the most optimal fine-grained controls -- this gives significant utility boost.