On July 5, 2023, OpenAI announced a new initiative, Superalignment, focused on achieving scientific and technical breakthroughs to steer and control potentially superintelligent AI systems. OpenAI created a new team, co-led by Ilya Sutskever and Jan Leike, and pledged 20% of its compute to the effort. OpenAI said:

Our goal is to solve the core technical challenges of superintelligence alignment in four years.

The core idea of OpenAI's approach is to "build a roughly human-level automated alignment researcher" and then to "use vast amounts of computer to scale our efforts, and iteratively align superintelligence."