OpenAI CEO Sam Altman concerned about artificial intelligence taking control over humans, ‘AI must always serve people’

Artificial intelligence (AI) should help people, but its rapid growth also brings serious risks. OpenAI chief Sam Altman has warned about two dangers that must be avoided. In a post on X, he explained why stronger safety checks are needed.

His first concern is that humans could lose control over powerful AI systems.

“This is unacceptable; we are unapologetically on Team Humanity, and AI must always serve people,” he wrote.

According to Sam Altman, this means safety measures must improve faster than these systems' capabilities.

His second concern is the concentration of power in too few hands across the world. One company, person or country could use powerful AI to force views on everyone.

“If an extraordinarily powerful AI is used by one person or company to impress their worldview onto everyone else, the results could be extremely dystopian,” Altman wrote.

According to the OpenAI CEO, leading AI companies must show the public that they are acting responsibly. He welcomed the US government's common safety rules for the most advanced AI.

Companies should build trust

However, he said companies should start building public trust without waiting for new laws. He also supported independent checks to examine whether companies are managing risks properly.

Earlier safety plans mainly focused on checking finished AI systems before their public release. Now, he argued, safety must receive attention while these systems are still being developed.

According to Altman, OpenAI prepares safety arguments before certain training steps expected to increase abilities. These arguments explain why the planned work can proceed safely, alongside existing release checks. He also urged companies to share safety methods and develop common standards together.

Altman clarified that controlling the speed of progress would not mean stopping AI development. Instead, companies should accept additional time and costs for testing, monitoring and safety. Pressure to compete, he said, cannot justify careless decisions about such powerful technology.

“Today's shift to focusing on safe development and evaluation will need new tools. For example, at OpenAI, we now formulate explicit safety cases in advance of frontier reinforcement learning runs we expect to significantly increase capability, in addition to the safety work we have long done in advance of model releases,” Sam Altman wrote.

“...no amount of American competitive pressure should justify recklessness, or let capabilities get ahead of alignment and monitoring,” he added.

“Where we will need the help of our government is for international coordination. But first we should do what we can ourselves,” he concluded.