The race to develop artificial intelligence (AI) is spiraling out of control, causing concern among people worldwide. However, three of the world’s leading figures in AI have reached a rare consensus: the rapid pace of AI development needs to be slowed down or limited for the time being. Following a cautionary blog post by Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman and xAI head Elon Musk also expressed their agreement via posts on the social media platform X (formerly Twitter). NDTV Online reported this development.
Elon Musk, the world’s wealthiest person, wrote on X that Dario was right. Meanwhile, Sam Altman stated, “I agree with Dario.” Altman also mentioned that his company would welcome third-party safety evaluators, in line with Amodei’s suggestion.
Amid growing concerns regarding the risks posed by “superintelligent” computer systems, Amodei wrote on his personal website that the pace at which AI models are gaining capabilities must be slowed down. He noted that while progress would still appear rapid, the time gained must be utilized wisely. He further explained that failing to develop the technology would deprive humanity of its benefits or allow it to fall into the hands of authoritarian powers, yet developing it too quickly would be reckless. While they had sought a middle ground, he realized in recent months that greater caution was needed to address the risks. Researcher Jacob Steinhardt (often referred to as Jacob Coxon in some reports) recently left the AI industry after raising allegations against both companies.
He had alleged that companies were “gambling” with human lives in the race to create self-improving models. Amodei’s warning came shortly after these allegations surfaced. Amodei warned that the actions of AI agents capable of operating without human oversight are not isolated incidents. Given the rapid growth of their capabilities, such harmful AI systems could seize control of the entire internet infrastructure within the next 6 to 12 months, potentially causing billions of dollars in damage. To mitigate this risk, Anthropic has pledged to grant independent third-party teams the same level of access as employees for internal safety evaluations. Additionally, the company’s head has urged all organizations to adhere to similar safety protocols.
