The mathematicians who raised $20 million to protect the world's largest AI labs
The startup Attestable has raised $20 million in a seed round to develop AI verification technologies. The team of mathematicians is building tools that allow for proving the correctness of computations without exposing sensitive data.
The company Attestable has revealed a $20 million seed round led by TLV Partners and Altimeter Capital, with participation from Netz Capital, Cerca Partners, and private investors including Wiz founder Assaf Rappaport, Armis founders Yevgeny Dibrov and Nadir Izrael, Sharin Fisher Dibrov, and Charlie Songhurst, a board member at Meta.
Attestable was founded in 2025 by three mathematicians: Yogev Bar-On (CEO), Shahar Papini (CTO), and Shahar Samucha (VP R&D). Bar-On worked at the RAND Corporation in the US as a policy advisor on AI security, where he was exposed to the difficulty of proving what happens inside AI labs. Papini and Samucha came from StarkWare, which specializes in zero-knowledge proofs and also participated in the funding round. Afterward, Papini worked at Ilya Sutskever's SSI, which he left to found Attestable.
The company employs 15 people, most of them at the development center in Israel and the rest in California, and will use the investment funds to expand the company's team and continue product development. Attestable is developing verification technology for AI systems that allows organizations to prove which calculations are being run and on what data, without exposing the model or the sensitive information itself.
The capability is intended for organizations running AI models for critical processes where security, resilience, and reliability are key conditions. The need for such a capability grows as AI systems move from experimental uses to decision-making and data processing in fields like finance and security. Organizations are required to have a high level of confidence in how the system operates, but running AI models involves a large number of components and complex systems, making them very difficult to monitor.
Attestable's technology is based on zero-knowledge proofs, a field of cryptography that allows one to prove that a certain claim is true without revealing the information used to prove it. The company has reached a mathematical breakthrough that allows this technology to be applied to AI systems to generate proofs that are quickly verifiable regarding how the model is operated, while maintaining the confidentiality of the data and the system.
The company aims to create the central tool through which the largest AI labs will protect their systems from hostile actors such as rival nations, and through which nations and large entities will be able to prove to each other that they are not using the technology for harm. In a conversation with Calcalist, Bar-On said:
«We are developing a way to secure AI labs like Anthropic and OpenAI against attacks by foreign adversaries. We are doing something different from Irregular — they do attack simulations, and we are building a system that allows labs to maintain security and prove to a third party that they are maintaining security. This allows several parties to coordinate what is happening in their organization and also prove to a second party what you are doing for security. We are a group of mathematicians who think we can solve the problem, and we managed to reach a significant breakthrough in the world of zero-knowledge proofs and practical technology for AI labs. Most were skeptical about our ability to reach this.»
Bar-On hopes that everyone running AI in the world will use the company's technology: «In a world of superintelligence, it is important to have control and verification mechanisms. In my vision, this will be an essential technology and part of regulation and agreements between nations. AI can be very good and cure cancer, but it can also do very bad things, and we must ensure it only does good. With our technology, we will be able to ensure that we only do good.»
According to Bar-On, «We don't have many expenses, we don't need much computing power. Our approach is a small but the best team there is. We recruited the best people in our field. Our solution is a bit expensive to use, but it's worth it.»
Yonatan Mandelbaum of the TLV fund told Calcalist that the negotiations for the investment took shape at a baseball game: «We sat at a game and five days later an agreement was reached.» According to him, «We understood that security for AI is interesting and important, and there are many good companies, but we looked at the short term — what is needed to deal with the technology today and what will come. A new paradigm is required in a world where AI will do everything, and a new firewall or new intrusion prevention software will not be enough. This new world is not far off. We saw the recently published attacks on the various models.»
«When I met Yogev and Shahar, I saw a new head for thinking about AI security,» added Mandelbaum. «This is not a security company, but a special company that has one of the strongest teams in the world in this field.»
