Humanity still has choice in frontier AI’s future, says Ex-OpenAI CTO

Mira Murati, former Chief Technology Officer of OpenAI, countered widespread pessimism about frontier AI’s future, insisting that humanity still holds agency in its development. Speaking to Emily Chang at Bloomberg Tech, Murati rejected simplified predictions of either utopia or dystopia, stating that the outcome depends on human choices in building and deploying AI tools. She acknowledged the potential risks and benefits of frontier AI systems, emphasizing the need for careful stewardship. Anthropic, a leading AI research lab, has raised alarms about self-evolving AI capabilities, warning that models like Claude may soon improve code 52 times faster than humans and potentially develop successors independently. The company called for a global pause in AI development, arguing that current advancements threaten to outpace human control. Murati acknowledged these concerns but framed them as opportunities for collaboration rather than inevitabilities. She stressed the importance of reducing ‘discontinuity’—sudden, unmanageable leaps in AI capability—that could destabilize human-AI partnerships. Murati suggested that proactive measures, such as gradual deployment and ethical safeguards, could mitigate risks while preserving AI’s transformative potential. Her perspective contrasts with Anthropic’s call for an immediate freeze, instead advocating for structured progress. The debate highlights tensions within the AI industry over speed, regulation, and ethical oversight. While some advocate for caution, Murati’s remarks reflect a belief that human ingenuity can guide AI’s evolution responsibly. The discussion underscores the need for balanced approaches as frontier AI continues to advance.