(Credit: Shutterstock / Likoper)
As AI-powered research surges, more than 100 scientists from universities and organizations around the world have agreed to adhere to 10 policies on AI research and development.
The agreement, crafted following a 20230 University of Washington summit and published on Friday, doesn't ban or condemn AI use. Rather, it argues that researchers shouldn't develop dangerous bioweapons using AI. Such an ask might seem like common sense, but the agreement details guiding principles that could help prevent an accidental DNA disaster.
"A new proactive risk management approach may be required to mitigate the potential of developing AI technologies that could be misused, intentionally or otherwise, to cause harm," the effort's website states.
The agreed-to policies state that scientists won't do research that could likely lead to "overall harm" or misuse, and will only source DNA services from those with screening measures that "detect hazardous biomolecules before they can be manufactured," among other precautions.
The commitments also state that scientists should determine safety risks "prior to release" of AI-powered software and participate in "red team" practices to look for potential issues in screening protocols.
Researchers at US institutions like Columbia, Harvard, Yale, MIT, Stanford, and many others have individually signed on to the agreement, as have scientists at universities in Canada, the UK, China, Germany, Israel, and Japan, to name a few. But these signatures only mean individual scientists—not the places where they work as a whole—are promising to adhere to the practices, according to the agreement's signatory page. It's possible, however, that some leaders may encourage their institutions and colleagues to adopt the policies as well.
On the corporate side, Microsoft's Chief Scientific Officer Eric Horvitz signed the agreement, and individuals at firms like ProteinQure, Menten AI, and Profluent Bio are also on the list.
Those behind the principles say that while AI poses unique risks, it will be a net-positive for the field of biology and humanity.
"Protein design is just the first step in making synthetic proteins," David Baker, director at the University of Washington's Institute for Protein Design, tells The New York Times, which first reported the news. “You then have to actually synthesize DNA and move the design from the computer into the real world—and that is the appropriate place to regulate."
But University of Chicago Professor Rama Ranganathan, who studies biochemistry and molecular biology and signed the agreement, says the scientific community will still be able to "freely explore" using AI in their fields despite the stipulations, according to the Times, which previously detailed how scientists are using AI tools to design new biological proteins that could fight illness.
This new optional agreement comes after Anthropic CEO Dario Amodei said last year that AI being used to develop bioweapons was a "medium-term" risk but could be just a few years away.
"Certain steps in bioweapons production involve knowledge that can't be found on Google or in textbooks and requires a high level of expertise," Amodei said. "We found that today's AI tools can fill in some of these steps."
While there aren't yet comprehensive US federal laws around the use of AI, the White House proposed an AI Bill of Rights blueprint back in 2022, and President Biden issued an executive order last year that calls for oversight of AI models that might threaten "national security, national economic security, or national public health and safety." As part of that EO, more than 200 companies and institutions joined the administration's AI Safety Institute Consortium earlier this year, which aims to tackle AI regulation and ensure the US stays at "the front of the pack."
Last year, 18 states and Puerto Rico approved various laws or resolutions surrounding the use of AI, the US National Conference of State Legislatures reports.