Former OpenAI CTO Mira Murati has announced her new startup. Unsurprisingly, it’s focused on AI.
Called Thinking Machines Lab, the startup, which came out of stealth today, intends to build tooling to “make AI work for [people’s] unique needs and goals,” and to create AI systems that are “more widely understood, customizable, and generally capable” than those currently available.
Murati is heading up Thinking Machines Lab as CEO. OpenAI co-founder John Schulman is the company’s chief scientist, and Barret Zoph, OpenAI’s ex-chief research officer, is the CTO.
In a blog post shared with TechCrunch, Thinking Machines Lab wrote that while AI capabilities have advanced dramatically, “key gaps” remain.
“The scientific community’s understanding of frontier AI systems lags behind rapidly advancing capabilities,” the blog post reads. “Knowledge of how these systems are trained is concentrated within the top research labs, limiting both the public discourse on AI and people’s abilities to use AI effectively. And, despite their potential, these systems remain difficult for people to customize to their specific needs and values.”
Thinking Machines Lab plans to focus on building “multimodal” systems that “work with people collaboratively,” according to the blog post, and that can “adapt to the full spectrum of human expertise and enable a broader spectrum” of applications.
“[W]e are building models at the frontier of capabilities in domains like science and programming,” the blog post stated. “Ultimately, the most advanced models will unlock the most transformative applications and benefits, such as enabling novel scientific discoveries and engineering breakthroughs.”
AI safety will be another core tenet of Thinking Machines Lab’s work. The company said that it plans to contribute to safety by preventing misuse of the models it releases, sharing best practices and recipes for how to build safe AI systems with the industry, and supporting external research on alignment by sharing code, datasets, and model specifications.
“We’ll focus on understanding how our systems create genuine value in the real world,” Thinking Machines Lab wrote in its blog post. “The most important breakthroughs often come from rethinking our objectives, not just optimizing existing metrics.”
I started Thinking Machines Lab alongside a remarkable team of scientists, engineers, and builders. We’re building three things:
– Helping people adapt AI systems to work for their specific needs
– Developing strong foundations to build more capable AI systems
– Fostering a…— Mira Murati (@miramurati) February 18, 2025
Murati left OpenAI last October after six years at the company. At the time, she said she was stepping away to “do her own exploration.”
Murati came to OpenAI in 2018 as VP of applied AI and partnerships. After being promoted to CTO in 2022, she led the company’s work on ChatGPT, the text-to-image AI DALL-E, and the code-generating system Codex, which powered early versions of GitHub’s Copilot programming assistant.
Mirati was briefly OpenAI’s interim CEO after CEO Sam Altman’s abrupt firing. Altman has described her as a close ally.
For months, rumors have flown of Murati hiring high-profile AI researchers and staffers for an AI venture. Thinking Machines Lab’s blog lists 29 employees from OpenAI, Character AI, and Google DeepMind, among other top firms.
Thinking Machines Lab is actively hiring machine learning scientists and engineers, as well as a research program manager, per the company’s post.
At one point, Murati was said to be in talks to raise over $100 million from unnamed VC firms. The blog didn’t confirm or deny this.
Before OpenAI, Murati spent three years at Tesla as a senior product manager of the Model X, the automaker’s crossover SUV, during which Tesla released early versions of Autopilot, its AI-enabled driver-assistance software. She also was VP of product and engineering at Leap Motion, a startup building hand- and finger-tracking motion sensors for PCs.
Murati joins a growing list of former OpenAI execs launching startups, including rivals such as Ilya Sutskever’s Safe Superintelligence and Anthropic.
