The leaders of prominent artificial intelligence firms are calling on Congressional members to enact legislation aimed at preventing the misuse of their technologies in the creation of biological weapons.
Among those advocating for these measures are Demis Hassabis from Google DeepMind, Sam Altman of OpenAI, Dario Amodei from Anthropic, and Mustafa Suleyman representing Microsoft AI. They have signed a public letter urging the enforcement of stringent regulations for companies engaged in the sale of synthetic DNA and RNA, emphasizing the need for meticulous customer screening to mitigate potential abuses of genetic resources.
The initiative was spearheaded by the nonpartisan Institute for Progress in collaboration with the right-leaning Foundation for American Innovation. The letter expresses concern over the rapid advancements in AI technology, warning that this progress could significantly lower the barriers that have historically kept malevolent actors from acquiring biological weaponry.
The journey of synthetic DNA began in the 1950s when scientist Arthur Kornberg successfully synthesized DNA for the first time. Today, this complex process has been automated, with many companies globally leveraging commercial synthesizers to manufacture and sell tailored genetic sequences utilized in various fields, including scientific research, drug development, and diagnostics. While numerous providers restrict sales to qualified researchers, biotech firms, and educational institutions, not all of them conduct thorough vetting of their customers or the genetic materials ordered.
Alarm over the potential ramifications of synthetic biology was notably raised in 2017 when Canadian researchers used $100,000 worth of mail-ordered DNA to recreate the long-extinct horsepox virus. Critics warned that the same techniques could potentially facilitate the construction of smallpox, a related and highly lethal virus. The cost of gene synthesis has only decreased since that incident.
In combination with AI enhancements, it has become increasingly viable to engineer dangerous toxins and pathogens using advanced language models, although some foundational biological expertise would still be necessary to create a functional virus from the ground up. Although instances of bioterrorism are relatively uncommon, their potential to inflict mass casualties, stir public anxiety, and result in significant economic disruptions poses a major risk. Of particular concern is the possibility that an AI-generated pathogen might unintentionally or deliberately trigger a global pandemic.
