Sam Altman, the CEO of OpenAI, has said the world should accept some harms caused by artificial intelligence in exchange for the technology's benefits. The statement, which he made without specifying what level of harm would be tolerable, puts him at odds with critics who argue that even limited damage from AI systems is too high a price. It also lands as governments and regulators are still working out how to oversee the fast-moving sector.
What Altman actually said
Altman framed the issue as a trade-off rather than a problem to be eliminated. He did not name specific incidents, nor did he put a number on what counts as an acceptable level of harm. That vagueness is likely to be a sticking point. Without a threshold, it's hard for regulators to write rules that distinguish between tolerable side effects and failures serious enough to require intervention.
His position isn't unusual for someone running a company that ships AI products at scale. Every deployed system carries some risk of error, bias, or misuse. The question is where the line sits, and who gets to draw it. Altman's answer — that some harm is worth it — is a broad principle, not a policy.
Why the comment matters for regulation
Altman's stance could shape how regulators approach AI oversight. If the head of one of the most prominent AI companies argues that a certain amount of harm is inevitable, it gives cover to lawmakers who prefer light-touch rules. It also gives ammunition to those who want stricter limits, because it confirms that harms are expected rather than accidental.
Regulators in multiple jurisdictions are still in the early stages of drafting AI rules. Some are focusing on transparency, others on safety testing, and others on liability. A public acceptance of harm from a major industry figure could tilt the debate toward frameworks that tolerate a baseline of negative outcomes, provided the overall benefits are judged to outweigh them.
The trouble is that "benefits" and "harms" are not evenly distributed. A model that improves productivity for many users can still cause concentrated damage for a smaller group. Altman's framing doesn't address that asymmetry.
Impact on industry standards
Altman's comments could influence industry standards, particularly if other AI developers follow his lead. If a major player says some harm is acceptable, it becomes harder for competitors to argue for zero-tolerance approaches. Standards bodies and trade groups often look to the largest companies for cues, and Altman's statement provides one.
That doesn't mean the industry will adopt his view wholesale. Companies selling into regulated sectors, such as health care or finance, may resist any framing that suggests errors are tolerable. But for general-purpose AI tools, the bar could shift.
Public trust and the unanswered question
Altman's stance could also affect public trust. Users who already worry about AI's impact may see the comment as confirmation that companies are willing to accept collateral damage. That perception can harden opposition and make it harder for regulators to justify a permissive approach.
There's no sign yet of how Altman's position will translate into OpenAI's actual policies or into the rules now being written. What's clear is that he has put a marker down: some harm is acceptable. The next step is for regulators and the public to decide whether they agree, and if so, how much.
