The generation of pathogen agents with pandemic potential using artificial intelligence (AI) is no longer a pipe dream. On August 6, the journal Science published a study describing the first viruses whose genomes were designed using AI tools developed by a team from Stanford University (California). Even though these harmless phages for humans were intended to attack bacteria with therapeutic aims, some scientists were alarmed by the announcement of these results, as the same tools could help create pathogenic agents.
Their fears were reinforced by a report released on September 10 by Anthropic, one of the American giants in AI. This report describes "five case studies of actors using AI agents in a way that promotes the development of biological weapons." Following this report, Anthropic's president, Dario Amodei, published proposals to slow down the race to develop the most advanced models, which could be diverted for harmful purposes, or become autonomous to the extent that their creators lose control.
He proposes an agreement between the United States and China, the two most advanced nations in this field, to "prohibit certain narrow and clearly dangerous uses of AI, such as using it to produce biological weapons or allowing users to do so." The two rivals should also agree to test their AI models before putting them on the market, "in view of the acute risks in cybersecurity, biology, and alignment [with human interests and values]."
You have 87.74% of this article left to read. The rest is reserved for subscribers.