OpenAI is an Anthropic Charter to Avoid Artificial Weapons of AI

CEOs a several major artificial intelligence companies are urging members of Congress to pass new legislation that would make it possible for bad guys to build weapons using their technology.

Demis Hassabis of Google DeepMind, Sam Altman of OpenAI, Dario Amodei of Anthropic, and Mustafa Suleyman of Microsoft AI are among the signatories. public letter calling for laws requiring companies that sell synthetic DNA and RNA to screen customers and mandates to prevent the misuse of genes.

Written by the nonpartisan Institute for Progress and the right-leaning Foundation for American Innovation, the letter acknowledges that due to the rapid development of AI, “there is a real possibility that the information barriers that have prevented criminals from gaining access to weapons of mass destruction will break down.”

Scientist Arthur Kornberg was the first to successfully synthesize DNA in the 1950s. Now, the process is automated, with many companies around the world using commercial technology to “print” and sell genetics used in scientific research, drug development, and disease. Many sponsors only sell to qualified researchers, biotech companies, and academic institutions, but not all of them buy from customers or the genes they order.

In 2017, Canadian researchers raised the alarm when he used the DNA of a $100,000 mailer to recreate the horsepox virus. Critics said the same method could be used to create smallpox, a highly related and deadly virus. Gene synthesis has only gotten cheaper since then.

Combined with advances in AI, it is possible to create new deadly toxins and pathogens using a large variety of languages, although some biology studies would still be needed to create a disease-causing virus from scratch. Although bioterror attacks are rare, they can cause mass casualties, public panic, and economic damage. A major concern is that an AI-engineered virus could intentionally or unintentionally cause a global pandemic.

“AI tools help the user to quickly identify where to turn to order things that cannot be measured,” said David Relman, an environmental and security expert at Stanford University, who signed the letter. “If asked properly, they can even tell you how to change your plan, so even the auditors can’t figure out what you’re trying to create.”

Signatories include other scientists, national security experts, and executives from genetic engineering companies Twist Bioscience and Ansa Biotechnologies. These companies are members of the International Gene Synthesis Consortium, which was founded in 2009 to use voluntary analysis. Many companies already use software to identify programs for “concerned effects” that may cause an organism to be endangered or cause disease.

“If you have technology that can process DNA, then you need to make sure it’s being used properly, and part of that is making sure you understand what you’re making and who you’re making,” says James Diggans, vice president of policy and biosecurity at Twist Bioscience. The company has helped to establish sustainable laws for many years.

Federal instructions provisions enacted during the Biden administration required scientists and companies receiving federal funding to order artificial genes from suppliers. A bipartisan bill passed earlier this year in the Senate would require all gene manufacturers operating in the U.S. to disclose to customers whether they are dealing with harmful or harmful pathogens.

But analysis tools are not perfect. Last year, Microsoft researchers published a learning showing that AI protein synthesis tools were able to create dangerous genes that contradicted previous industrial programs. The samples yielded new protein structures with similar properties to those known to be toxic.

Geoff Ralston, former president of Y Combinator and a partner at the Safe AI Fund, thinks that AI laboratories with biology models should test their users themselves.

Ralston, who also signed the letter, said: “It must be very difficult, if not impossible, to ask an imaginary person to help you do something very dangerous.

Relman acknowledges that regulatory frameworks are only part of the answer. “Since monitoring can fail in some cases, we need to have some control measures in place,” he says. “That’s where the AI ​​industry needs to step up.”