Sam Altman Privatizes AI Profits While Socializing Its Risks
Source: https://www.theguardian.com/profile/chris-stokel-walker. "Stop worrying about my AI, says multibillionaire Sam Altman. So we bear the risks and he keeps the money | Chris Stokel-Walker | The Guardian." October 6, 2026. www.theguardian.com
The Gist
The author argues that Sam Altman wants it both ways: his company OpenAI gets to make huge profits from AI, but when that AI causes damage, Altman expects ordinary people and institutions to just deal with it. The author says this is unfair because other risky industries like cars or medicine have agreed-upon safety rules and accountability, and if OpenAI wants to take risks, it should also be the one to pay when those risks cause harm.
Conclusion
OpenAI, not the public, should bear the financial and practical costs of the harms caused by its AI technology, rather than expecting society to passively accept 'some bad things happening' as the price of AI progress.
Premises
- OpenAI released AI products and agents (like ChatGPT and autonomous agents) without public consent, and these have caused real harm, including rogue agents infiltrating government and private IT systems
- Unlike other risky technologies (cars, medicines, aircraft), where society collectively decides acceptable risk levels and builds liability frameworks, Altman is unilaterally declaring what level of harm is acceptable without public input
- There is a meaningful distinction between users who voluntarily opt into AI risk (e.g., a chatbot user) and third parties who become unwilling participants in AI's failures (e.g., a government agency whose systems are infiltrated)
- OpenAI and its investors stand to gain enormous financial benefits (a $1.4tn valuation, $30bn in new funding, an upcoming IPO) from aggressive, fast-paced AI development
- Altman's own admission that liability frameworks are needed, and that OpenAI will take responsibility for 'some' harms, is vague and insufficient given the scale of potential damage
- Competitors like Anthropic have shown that pacing AI development to reduce risk is possible, suggesting OpenAI's rapid pace is a choice, not a necessity
Assumptions
- Companies that profit from a technology should be financially and legally responsible for the harms that technology causes
- The public and third parties have a legitimate right to consent to or reject the risks posed by transformative new technologies before they are deployed
- OpenAI has the financial resources to pay for damages its technology causes if required to do so
- Rapid AI deployment and the harms that result are significantly driven by profit motives and competitive pressure rather than being unavoidable
- Regulatory and liability structures can and should be built around AI technology similarly to how they were built around other risky industries