Anthropic CEO proposes framework to slow artificial intelligence development

Dario Amodei, the CEO of AI company Anthropic, published an essay on Saturday calling on AI companies to slow down their development of new systems. He warned that AI could become dangerous within six to twelve months, when it may be able to attack internet security on a large scale. He proposed a three-step plan to delay development and give researchers more time to manage risks. He estimated that a slowdown could buy one to two years before AI reaches a critical level. Anthropic has already taken the first step, giving outside evaluators permanent access to check its systems. Amodei said any slowdown must happen internationally. Sam Altman of OpenAI and Elon Musk of xAI both said they agreed with him.