OpenAI CEO Warns of AI Superintelligence Risks, Calls for U.S. Preparedness

OpenAI CEO Sam Altman has called on U.S. policymakers to take immediate action to address the growing risks and opportunities posed by advanced artificial intelligence. In an interview with Axios, Altman emphasized that AI is rapidly transitioning from theoretical concepts to practical applications in the economy, with systems already performing tasks such as coding and research that previously required large teams of experts. He warned that the pace of development could lead to significant disruptions, particularly in cybersecurity and other critical sectors. The shift is already evident in the cybersecurity landscape, where AI tools are reducing the time and expertise needed to identify and exploit software vulnerabilities. Charles Guillemet, CTO of hardware wallet maker Ledger, noted that tasks like reverse-engineering code or linking vulnerabilities, which once took months, can now be accomplished in seconds using AI. This has implications for the crypto industry, which experienced over $1.4 billion in stolen or lost assets in attacks last year. Guillemet suggested that these figures could rise as AI-generated code becomes more prevalent, potentially introducing new flaws at scale. Altman also highlighted the dual-use nature of AI, noting that while it could accelerate breakthroughs in fields like drug discovery and materials science, it could also enable more sophisticated cyberattacks and lower the barrier to harmful biological research. He stressed the urgency of coordination between government, tech firms, and security groups to mitigate these risks. 'We’re not that far away from a world where there are incredibly capable open-source models that are very good at biology,' Altman said, warning of the potential for terrorist groups to use such models to create novel pathogens. In addition to the technical and security challenges, Altman addressed the economic and societal impacts of AI. He described a future where AI becomes a utility, much like electricity, with costs decreasing for basic applications while high-end systems remain expensive. This shift could lead to significant labor changes, with programmers and other professionals adapting to new workflows. Altman also mentioned the importance of ensuring that AI developers are of high integrity and trustworthy, given the potential for misuse. The CEO also commented on the potential nationalization of OpenAI, arguing that the U.S. must achieve 'superintelligence' before its rivals to maintain a competitive edge. He suggested that a government-led project might not be effective, but emphasized the need for close collaboration between private companies and the U.S. government to manage emerging risks effectively.