In a recent article, Steven Levy explores the ambitious goals of Anthropic, co-founded by siblings Daniela and Dario Amodei, who previously worked at OpenAI. Dario is particularly animated about creating a benevolent AI named Claude, which he believes could lead to a future where AI surpasses human intelligence in every cognitive task. Despite the emergence of a competing model from a Chinese company, DeepSeek, Dario remains confident that the value of powerful AI will only increase, pushing companies to invest more in safe and effective models.

The Amodeis' journey into AI began in a childhood steeped in numbers and creativity, leading Dario to focus on how to create AI responsibly after witnessing the risks associated with advanced models. Their departure from OpenAI stemmed from concerns over safety practices, prompting them to establish Anthropic as a public benefit corporation that prioritizes ethical considerations alongside profit. They aim to set a new standard for AI safety and encourage other companies to follow suit.

Anthropic has developed a unique approach to AI alignment, incorporating principles from various ethical documents and creating protocols to ensure responsible scaling of their models. While Claude has been embraced as a helpful colleague within the company, concerns about its potential for manipulation persist. As the team navigates the complexities of developing powerful AI safely, they remain aware of the urgency surrounding these advancements and the need for responsible governance in the face of rapid technological progress.