OpenAI co-founder wants to create "secure superintelligence"

OpenAI co-founder Ilya Sutskever, who left the artificial intelligence (AI) company OpenAI, wants to create safe superintelligence (SSI) in his new company Safe Superintelligence Inc. The newly founded start-up does not yet have much to show.

A simple website with a reference to Safe Superintelligence, social media accounts and a small team are all that point to Sutskever's project so far. "Superintelligence is within reach. Building secure superintelligence (SSI) is the most important technical problem of our time," reads the website. The issue of security is a constant source of contention at OpenAI. Due to differing views, Sutskever and the head of the security team, Jan Leike, have left the company at their own request.

Secure superintelligence

A small team of around three people will initially dedicate themselves to the goal of a secure AGI. In addition to Sutskever, these are Daniel Groß, former head of AI at Apple, and Daniel Levy, an ex-employee of ChatGPT creator OpenAI. The three have signed the announcement on the website. According to the company, further employees will be hired to build "the world's first SSI lab". The "best engineers and researchers in the world" are to form a "lean, powerful team", it continues. The employees will work at the Palo Alto site in the US state of California and in Tel Aviv, Israel.

Sutskever and his colleagues intend to create a super-intelligence that poses no threat. This is the company's sole aim. There is to be no other product. The safety of the superintelligence is to be guaranteed by placing a special focus on it during development. Security is the top priority, the signatories emphasize on the website. They do not want to be subject to any commercial pressure. Safety takes precedence over product cycles.

Sutskever and his fellow campaigners do not provide any details about the SSI. What it will do and when it might come is uncertain. The financing is just as unclear. The announcement merely states that investors are available. The young start-up did not say who these investors are or what the AI company's financial resources look like.

Sutskever is considered one of the world's leading minds in artificial intelligence. The Canadian computer scientist co-founded OpenAI and played a key role in the development of ChatGPT. Sutskever left OpenAI in May 2024 after ten years with the company because he feared that commercial interests could take precedence over security in the further development of ChatGPT. Back in May, Sutskever announced a project that was close to his heart.

(olb)