Igor Babuschkin, a co-founder of Elon Musk's AI startup xAI, is stepping down to launch his own fund focused on AI safety. His departure comes as xAI faces criticism over controversies surrounding its chatbot Grok.
In a post on X, Babuschkin announced he is leaving xAI to start "Babuschkin Ventures," a venture capital fund aimed at supporting startups working on AI safety and agentic systems.
At xAI, Babuschkin was responsible for engineering, infrastructure, and applied AI projects. He says he developed many of the early tools needed for training processes and highlights his role in building the "Memphis Supercluster" - a supercomputer for AI training that he claims was assembled in just 120 days. According to Babuschkin, the project only came together through intense effort and strong team spirit.
Focus on AI safety after xAI controversies
Babuschkin says he now wants to dedicate himself to building safer, more human-centered AI. He points to a conversation with Max Tegmark, founder of the Future of Life Institute, as a key influence. Over dinner, Tegmark showed Babuschkin a photo of his two young sons and raised the question of how to design AI so that future generations can grow up in a safe and supportive environment. Tegmark is one of the leading researchers who called for a pause in AI training shortly after the release of GPT-4.
Babuschkin says the meeting deeply affected him and underscored the responsibility involved in shaping advanced AI systems. With Babuschkin Ventures, he plans to back research and startups focused on the safety and ethical challenges of intelligent, agentic AI - with the goal of contributing to positive societal outcomes.
His exit comes at a turbulent time for xAI. In recent months, the company has faced backlash after its chatbot Grok made several controversial statements. Most recently, a new Grok feature drew attention for letting users generate AI-created nude videos of public figures.
Babuschkin does not directly link his departure to these scandals, but the timing suggests that internal tensions and ethical concerns may have played a role. xAI has also faced criticism from safety researchers for not releasing model or system cards detailing its safety testing, unlike other companies in the field.
Before founding xAI, Babuschkin worked at OpenAI and DeepMind. At DeepMind, he served as a technical lead on the AlphaStar project, an AI system that defeated professional StarCraft players.
