A new RAND analysis warns that advanced artificial intelligence could make it easier for malicious actors to develop biological weapons, including scenarios involving biological events with impacts ten times greater than the COVID-19 pandemic, while researchers call for multiple overlapping safeguards.
By yourNEWS Media Newsroom
Artificial intelligence could significantly lower the barriers to creating biological weapons and eventually enable attacks far more destructive than past pandemics, according to a new RAND report examining how governments and technology companies could reduce the risks posed by rapidly advancing AI capabilities.
The RAND report, published this week, outlines nine potential interventions intended to make it more difficult for malicious actors to use artificial intelligence for dangerous biological research.
RAND_RRA4999-1 by yourNEWS Media
The warning comes as AI companies demonstrate increasingly sophisticated capabilities in fields such as protein design.
Anthropic announced Tuesday that its Claude model successfully designed proteins that bound to 14 of 15 intended targets during laboratory testing, according to the company.
Such advances have potential applications in medicine and biotechnology, but researchers and AI developers have also raised concerns that the same capabilities could be misused to assist in the development of biological weapons.
“AI-enabled biotechnology might lower the technical, operational, and motivational barriers to creating biological weapons,” RAND warned. “It might also enable devastating attacks involving enhanced or novel pathogens, including events far worse than historical pandemics that caused many millions of fatalities globally.”
RAND’s analysis specifically concentrates on extreme biological scenarios rather than predicting that such an event is imminent.
“[This report] focuses on high-consequence AI-enabled biological incidents, which we define as biological events capable of producing impacts on the order of ten times the scale of the coronavirus disease 2019 (COVID-19) pandemic or greater, societal disruption, or systemic destabilization,” the report’s Purpose section states.
The growing concern over biological misuse has already prompted leaders at major artificial intelligence companies to seek additional federal safeguards.
Executives associated with Google, OpenAI and Anthropic urged Congress in June to establish legal requirements for companies selling DNA and RNA.
The proposed protections would require sellers to screen customers and maintain records of orders, according to The Wall Street Journal.
The idea is to create another security barrier between AI-generated biological information and the physical materials that could be necessary to act on it.
RAND argues that no single intervention is likely to be sufficient.
“No single safeguard can prevent a bad actor from using AI to help build a biological weapon,” the organization wrote Wednesday on X. “But our new report identifies nine interventions that, layered together, could meaningfully lower the risk of a significant AI-enabled attack.”
That layered approach reflects the difficulty of controlling technology that can simultaneously support legitimate scientific research and potentially dangerous applications.
Anthropic has acknowledged that its work in protein design presents precisely that kind of dual-use challenge.
“Without robust safety measures, they could enable bad actors to perform dangerous research, such as the development of bioweapons,” the company said in its Tuesday announcement.
Anthropic said it is therefore withholding some advanced biological capabilities from general users while it develops systems for restricted access.
“As we work to deliver these capabilities safely via trusted access programs, protein design and other dual-use research biology capabilities remain unavailable for general access in Claude Fable 5,” the company said.
The laboratory results announced by Anthropic demonstrate how quickly AI systems are moving beyond text generation into tasks with direct relevance to biological research.
According to the company, Claude-generated protein designs successfully bound to 14 of 15 laboratory targets, illustrating the potential for AI to accelerate work that previously required substantially more specialized expertise and time.
That ability could benefit drug development and other legitimate scientific work. RAND’s concern is that similar advances may also reduce the level of knowledge, training and resources needed by someone seeking to pursue harmful biological research.
RAND lead author Steph Guerra said the most alarming capabilities envisioned in the report have not yet materialized.
The “most dangerous thresholds have not yet been crossed,” Guerra told Axios.
She nevertheless argued that the pace of technological development makes it important to establish safeguards before those thresholds are reached, calling for “mutually reinforcing approaches” rather than reliance on a single regulatory or technical solution.
Neither RAND nor Anthropic immediately responded to requests for comment from the Daily Caller News Foundation.
The report arrives as policymakers confront a broader challenge surrounding advanced AI: the same systems capable of accelerating medical and scientific discovery may also make highly specialized technical knowledge more accessible to people who previously lacked the expertise to use it.
RAND’s analysis does not conclude that AI has already made a catastrophic biological attack possible. Instead, it warns that continuing advances could progressively lower the barriers to such an event and argues that governments, biotechnology companies and AI developers should build multiple layers of protection before those capabilities become widely available.
Original article
Comments