San Francisco. After the shock of independent hacking attacks by artificial intelligence, major American AI companies say they are prepared to slow down the development of the technology. Dario Amodei, head of the AI company Anthropic, made such an advance with the plan to gain one to two years of time. Within a few hours he received support from rivals such as the head of the ChatGPT developer OpenAI, Sam Altman, and tech billionaire Elon Musk, who has his own AI developed at his company SpaceX.
After the hacking attacks, Amodei wrote on Saturday that he was concerned that in six to 12 months an AI swarm could be able to take over the entire Internet and potentially cause hundreds of billions of dollars in damage. OpenAI's Altman wrote on Platform
Amodei argued for a coordinated industry-wide pause in the Western world. At the same time, without waiting for others, Anthropic will, as a first step, allow permanent independent observers into the company to assess the safety of AI models during the training phase. Altman called this a good idea and announced that OpenAI would do the same. Tech billionaire Elon Musk, long a critic of Amodei, also wrote on X: “Dario is right.”
Altman also told Fortune magazine that OpenAI's highly anticipated IPO would probably not take place this year. He particularly referred to the current debate: “It would be ill-advised to go public right now.” When asked whether he could also imagine foregoing an IPO next year, Altman replied: "I would say not in 2026."
ChatGPT sparked the current hype about artificial intelligence several years ago. OpenAI is expected to not only raise tens of billions of dollars when it goes public, but also achieve a very high valuation. The share placement has been planned for some time.
For weeks there have been great concerns about the risks of artificial intelligence. In the recently reported sensational hacking attack, software from OpenAI broke out of a secure environment in a test run and penetrated the systems of the AI company HuggingFace on its own initiative. The only reason was that she suspected there would be answers to the test task given to her. For the attack, the AI agents exploited software vulnerabilities, among other things - and also coordinated with each other.
Amodei emphasized that it was worrying that an AI swarm with the same attitude and greater capabilities could have caused “catastrophic damage.” He sees the ability of AI software to further develop itself as a particular risk. Since this summer, artificial intelligence has been better at it - and this is increasing the pace even further. If left uncontrolled, this could lead to people no longer understanding these systems and losing control over them, he warned.
In recent years, Amodei has repeatedly warned about the risks associated with the development of artificial intelligence - and was sometimes heavily criticized for this by parts of the AI industry. One argument from supporters of a high pace in the USA - including in President Donald Trump's White House - is the race with China.
Anthropic recently developed an AI model called Mythos, which is particularly good at finding software vulnerabilities that have remained undetected for decades. The company makes the full model available only to selected authorities and companies so that they can secure their computer systems.
The debate had been fueled in recent days by an Anthropic researcher who resigned with a drastic warning. The major developers of artificial intelligence are acting irresponsibly “and playing with our lives,” explained Jacob Coxon.
Anthropic researcher Evan Hubinger, who is responsible for ensuring that AI stays in line with people's interests, confirmed Coxon's concerns a short time later. He personally thinks that the risk of AI killing humanity is more than ten percent in the next decade, he wrote on X.
There is also a growing movement in the USA against the huge data centers that AI companies like OpenAI rely on to build their businesses.
