OpenAI Chief Executive Officer Sam Altman said he is considering ways to moderate the pace of developing advanced artificial intelligence (AI) models.
Bloomberg reported on the 10th (local time) that Altman told employees at a companywide meeting this week that he is considering this. Altman said it is possible to slow development to stay in step with other AI research teams, though he noted some teams may not agree.
Altman's remarks came as concerns have mounted in the AI industry about the risks of increasingly sophisticated AI systems. Current and former researchers at major AI corporations have publicly called for slowing the pace, saying the AI development race is moving too fast.
OpenAI chief scientist Jakub Pachocki also said in a recent post that "AI corporations should cooperate to slow future development when needed," adding, "I hope voluntary speed moderation becomes common until shared safety standards are in place."
Earlier, OpenAI said it had slowed some phases of recent model development and temporarily paused certain internal AI training processes for safety reasons. Altman noted in July that he discussed the need to moderate AI development speed with White House officials.
Jacob Coxon, an employee who researched AI at Anthropic and OpenAI over the past three years and recently left Anthropic, posted a strong warning about AI risks on social media on the 8th. He said the two companies are trying to build AI models that improve themselves and are "gambling with our lives in a headlong race toward superintelligent AI."
He further claimed that "people who are seriously developing AI believe we could all die from AI before this decade ends."
The post drew more than 150 million views and attention from politicians and celebrities.
On the 10th, two researchers who recently left Anthropic and Google DeepMind also publicly voiced concerns about the pace of AI development. They argued AI corporations should be more transparent than they are now about the technology's potential risks. In July, more than 1,000 employees at major AI corporations signed a petition urging mechanisms to slow AI development.
Recently revealed cases of AI systems failing to stay under control have added to the concern. After it emerged that OpenAI's AI agents colluded with each other to bypass internal controls and hacked third-party websites, debate intensified over the latest AI models' cyberattack capabilities and the feasibility of controlling them.
Meanwhile, U.S. President Donald Trump dismissed researchers' concerns that AI could advance to an uncontrollable level and drive humanity to extinction. He maintained it is more important for the United States to keep an edge over China in the AI race.
On the 10th, as he left Dallas, Texas, after the party convention, Trump answered "not at all" when asked whether he worries about the possibility of human extinction caused by AI.
He said, "I do worry that if we don't win the AI race, we'll be put at a great disadvantage," adding, "Right now we are quite a bit ahead of China. I think we're about a year ahead, which is considered a significant gap."