OpenAI CEO Sam Altman said U.S. policymakers need to move quickly to prepare for more advanced artificial intelligence, arguing that the technology is no longer a distant concept and is already entering everyday economic activity. In his view, AI systems are already handling coding and research tasks that once required teams of programmers. Newer models will push that much further, he said, helping scientists make major discoveries and enabling individuals to do work that previously belonged to entire groups.
AI is changing the cost structure of cyberattacks
That shift is already visible in cybersecurity. Ledger Chief Technology Officer Charles Guillemet told CoinDesk that AI tools are lowering both the cost and the skill required to discover and exploit software flaws. Work that once took months, such as reverse-engineering code or chaining multiple vulnerabilities together, can now be completed in seconds with the right prompts. The pace is different. So is the barrier to entry.
The article notes that the crypto industry saw more than $1.4 billion in assets stolen or lost in attacks last year. Guillemet suggested that total could keep rising. He also pointed to a second pressure point: developers are using more AI-generated code, and that may introduce new flaws at scale. His proposed response focused on stronger defenses, including mathematically verified code, hardware devices that keep private keys offline, and broader acceptance that systems can fail.
Altman highlights biosecurity and large-scale cyber threats
Speaking to Axios, Altman said AI could accelerate drug discovery and materials science, but he also warned that it could strengthen cyberattacks and reduce the barrier to harmful biological research. He said such threats may emerge within a year, which makes coordination across government, technology companies and security groups urgent.
He said the world is not far from a point where highly capable open-source models are very good at biology. In that scenario, society would need to be resilient against terrorist groups trying to use those models to create novel pathogens. He also raised the prospect of a “world-shaking cyberattack” that could arrive as early as this year. Avoiding that outcome, he said, would require a tremendous amount of work.
OpenAI pushes collaboration with government, rejects nationalization case
Altman described OpenAI’s policy proposals as a starting point for debate over how to manage systems that learn quickly and operate across many domains. He also said AI itself should be used to help defend against these threats.
On the question of nationalizing OpenAI, Altman argued against it on the grounds that the U.S. needs to reach “superintelligence” before rivals do, and to do so in a way aligned with democratic values in the United States. He said that likely would not work as a government project, while adding that companies involved in AI still need to work closely with the U.S. government.
The report also noted that Altman has a financial stake in how the AI sector develops because of his role at OpenAI. That position may shape how he presents both the urgency of regulation and the role private companies should have in managing emerging risks, with possible implications for OpenAI’s competitive position.
AI as a utility with usage-based costs
Beyond security, Altman pointed to energy and labor as areas where AI’s effects are becoming clearer. He said energy may see rapid progress because greater processing power capacity could help keep costs down as AI demand rises. He also said labor patterns are already shifting, giving the example that a programmer in 2026 works differently from one just a year earlier.
His broader view is that AI will become a utility, comparable to electricity, embedded across devices as the cost of basic intelligence falls while top-tier systems remain expensive. He described a model in which a personal “super assistant” runs in the cloud, with monthly bills rising for heavier use or higher levels of intelligence and falling when usage is lighter.