The former Openaai CTO Mira Murati has announced her new startup. It is not surprising that it focuses on AI.
Called Memory laboratoryThe startup, which came from Stealth today, intends to create tools in order to make “AI work for (unique needs and goals of people) and to create AI systems that are“ generally understood, adaptable and generally capable ” these currently available.
Murati heads the memorial laboratory as CEO. Openai co-founder John Schulman Is the chief scientist of the company, and Barret ZophOpenas ex-chief research officer, is the CTO.
In a blog divided with techcrunch, Thinking Machines Lab wrote that the AI functions have dramatically developed, “important gaps” remain.
“Understanding the scientific community for Frontier -KI systems remains behind rapidly progressive skills,” says the blog post. “Knowing how these systems are trained focuses in the top research laboratories and limits both the public discourse via AI and the skills of people to use AI effectively. And despite their potential, these systems remain difficult for humans to adapt to their specific needs and values. “
Memorials Labin plans to concentrate on the structure of “multimodal” systems that, according to Blog post, “work with people” and adapt to the entire spectrum of human expertise and enable a wider spectrum of applications.
“(W) e build models at the boundary of skills in domains such as science and programming,” says the blog post. “Ultimately, the most advanced models will unlock the most transformative applications and advantages, such as new scientific discoveries and technical breakthroughs.”
The security of AI will be another core principle for memory laboratories. The company said that it should contribute to security by preventing the abuse of the models it published, best practices and recipes for building safe AI.
“We will concentrate on understanding how our systems in the real world create a real value,” wrote Thinking Machines Lab in his blog post. “The most important breakthroughs often come from the rethinking of our goals and not only from the optimization of existing metrics.”
I started thinking with a remarkable team of scientists, engineers and builders. We build three things:
– People help to adapt AI systems to work for their specific needs
– Development of strong foundations to build more powerful AI systems
– Promotion of a …– Mira Murati (@miramurati) February 18, 2025
Murati left Openai last October after six years at the Openai company. At that time, she said she was leading back to “investigate her own exploration”.
Murati was opened in 2018 as a VP of applied AI and partnerships. After being promoted to CTO in 2022, she directed the company’s work ChattThe text-to-image ki Out ofand the code-generating system CodeThe early versions of toses Githubs Copilot Programming assistant.
Mirati was short openais interim CEO to CEO Sam Altman’s abrupt shooting. Altman described it as a narrow ally.
For months, rumors have been flown over Murati who have hired top-class AI researchers and employees to obtain a AI company. Thinking Machines Lab’s blog lists 29 employees of Openai, Character AI and Google Deepmind.
Thinking Machines Labin Labor’s machine scientists and engineers for machine learning as well as a research program manager in accordance with the company’s post.
At some point Murati should be in talks to collect over 100 million US dollars from unnamed VC companies. The blog did not confirm or rejected this.
Before Openaai, Murati was spent three years at Tesla as Senior Product Manager of Model X, the car manufacturer’s crossover SUV, in which Tesla released early versions of Autopilot. It was also VP of product and engineering at Jump movementA start-up building handle and finger tracking movement sensors for PCs.
Murati joins a growing list of former Openai managers who start startups, including rivals such as Ilya Sutskevers safe superintelligence and anthropic.
