OpenAI Boss Says World ‘Right to Be Afraid’ but ‘Should Trust’ AI Firms

By Angel No Lie | KPD Online | Independent Technology Report | 17 September 2026

OpenAI chief executive Sam Altman has acknowledged that public concern about increasingly powerful artificial intelligence is justified, while arguing that people should nevertheless trust AI companies to act responsibly as the technology develops.

Speaking at Salesforce’s annual Dreamforce conference in San Francisco on September 15, Altman said AI had advanced to a point where the potential consequences of failures were becoming easier to imagine. He also identified another concern: the possibility that a small number of AI companies could accumulate excessive economic and social power.

“The world is right to be afraid of this,” Altman said, while also arguing that people should trust AI companies to make responsible decisions.

Two risks highlighted by Altman

Altman’s comments focused on two broad categories of risk.

The first is what he described as the possibility of a loss-of-control accident or another serious failure involving increasingly capable AI systems.

The second is concentration of power. Altman warned that AI companies could potentially acquire enough influence to affect the economy or push particular worldviews onto the public.

That distinction is significant because the AI safety debate is no longer limited to hypothetical questions about future superintelligence. It also includes more immediate questions about who controls advanced models, how they are deployed and what safeguards apply to them.

Technology Leaders Speak At Annual Dreamforce Event In San Francisco

Altman argues the industry can manage the risks

Despite acknowledging those dangers, Altman expressed confidence that AI companies can develop the technology safely.

He said companies should maintain safety and alignment ahead of their capabilities and indicated that development could be slowed or stopped if necessary.

The argument places considerable emphasis on responsible behaviour by the companies themselves rather than relying exclusively on government regulation.

That approach is becoming one of the central issues in the international AI debate.

Industry disagreement over regulation

Altman’s remarks came during a week of unusually intense discussion about AI safety.

Anthropic CEO Dario Amodei has called for a slower pace of AI development and stronger safeguards. Altman has expressed support for coordinated efforts to improve safety, while other technology executives have argued that companies can manage the risks internally without extensive new government regulation.

The debate therefore involves two related but different questions:

Calls for greater external oversight

Not everyone agrees that the technology companies should be trusted to police themselves.

The Associated Press reported that the recent debate has exposed divisions within the AI industry over whether there should be coordinated restrictions or a slowdown. Some AI leaders have advocated stronger external safeguards, while others maintain that companies have sufficient incentives and technical expertise to manage safety themselves.

Yoshua Bengio, one of the pioneers of modern AI, has argued for ambitious safety efforts outside the for-profit sector, according to AP.

That disagreement reflects a broader question of governance: can companies developing the most powerful AI systems simultaneously be the primary institutions responsible for determining how those systems should be controlled?

There is no universal agreement on the answer.

Why the debate has intensified

The latest discussion follows warnings from researchers and technology executives about the possibility that increasingly autonomous AI systems could create serious risks if safeguards fail.

At the same time, experts do not have a consensus on the probability or timing of catastrophic AI scenarios. AP reported that proposed risks range from malicious use of AI to hypothetical situations in which future systems become difficult for humans to control.

This uncertainty is important. Acknowledging that a risk is possible does not establish that it will happen, nor does it establish how likely a particular scenario is.

The trust question

Altman’s remarks put the issue of trust at the centre of the discussion.

AI companies argue that they have strong technical knowledge, commercial incentives and reputational reasons to prevent their products from causing serious harm. Meta CEO Mark Zuckerberg, for example, has argued that AI laboratories have both the ability and incentive to ensure their systems remain aligned with human values.

Critics counter that commercial competition can create pressure to release increasingly capable systems quickly, potentially creating a conflict between speed and safety. The debate therefore extends beyond whether individual executives are acting in good faith; it concerns what institutional safeguards should exist regardless of who is running a company.

What remains unresolved

Several major questions remain open:

• How much autonomy should advanced AI systems be permitted to have?
• What safety tests should be mandatory before powerful models are released?
• Should independent organizations have access to evaluate frontier AI systems?
• When should governments intervene?
• How should responsibility be assigned when AI causes harm?
• Can international safety standards keep pace with rapid technological development?

These questions are likely to remain central as AI systems become more capable and more deeply integrated into business, government and everyday life.

Independent assessment

Altman’s comments represent a notable acknowledgment from one of the industry’s most prominent executives that fear of AI is not inherently irrational. At the same time, his call for public trust highlights the unresolved tension between industry self-governance and external oversight.

The available evidence does not establish that AI companies can or cannot safely regulate themselves on their own. What is clear is that there is an active disagreement among technology leaders, researchers and policymakers over the appropriate balance between innovation, corporate responsibility and independent oversight.

For the public, the practical issue is therefore not simply whether to trust or fear AI. It is whether the systems governing AI development provide enough transparency, accountability, testing and independent scrutiny to justify that trust.

Sources: Reuters, Associated Press, Axios, WIRED and Salesforce Dreamforce coverage.