AI Leaders Rarely Agree On The Future Of The Technology. Many Share The Same Fear About Bioweapons
Artificial intelligence executives often disagree on regulation, open-weight models, and the pace at which new systems should be released. But one issue continues to draw broad agreement across the industry: the risk that advanced AI could someday be misused in biological research.
The discussion has become increasingly prominent as governments, technology companies and national security officials examine the capabilities of frontier AI models and the safeguards surrounding them. It has also gained attention amid growing geopolitical tensions and heightened concerns over biological preparedness in the years following the COVID-19 pandemic.
Many of the people building today’s leading AI systems privately identify engineered pathogens as their greatest concern, Axios reported Thursday. The publication said executives and researchers have repeatedly pointed to the possibility that future systems could assist bad actors in understanding biological vulnerabilities or accelerating parts of the research process.
A recent study from MIT FutureTech and the University of Queensland surveyed 272 researchers across 24 categories of AI risk. Participants estimated a 12% probability that dangerous AI capabilities could contribute to a catastrophic outcome by 2030 and a separate 12% probability that AI-enabled weapons or mass-harm capabilities could emerge during the same period. Researchers said those figures exceeded 20% in scenarios without mitigation efforts.
The study defined a catastrophic event as one involving more than one million deaths or at least $100 billion in damages. Risks involving chemical and biological weapons were among the highest-ranked categories identified by respondents.
The concerns are shared across competing companies. Last month, OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei, Google DeepMind CEO Demis Hassabis, Microsoft AI CEO Mustafa Suleyman and Meta Chief AI Officer Alexandr Wang signed an open letter calling for stronger protections against AI-assisted biological threats, according to The Verge. The letter urged lawmakers to require screening of synthetic DNA and RNA orders and improve safeguards around biological research.
The issue has become increasingly relevant in Washington. Congress has held multiple hearings on AI and national security, while agencies including the Department of Homeland Security have examined the implications of advanced AI systems for biosecurity and public health. Axios previously reported that DHS has identified biological threats as one of the most significant areas where AI presents both opportunities and risks.
At the same time, advances in biotechnology continue to reduce the cost and complexity of scientific research. Gene-editing tools such as CRISPR have become more widely available, while AI systems are increasingly used in drug discovery, protein analysis and genomic research.
Scientists have emphasized that creating a novel pathogen remains a difficult undertaking requiring specialized expertise, laboratory infrastructure and extensive experimentation. However, several researchers have argued that AI could accelerate parts of that process by helping identify promising biological candidates or synthesizing large volumes of scientific information. TIME reported last year that experts from institutions including Stanford and Johns Hopkins had warned that AI could lower barriers to bioterrorism if safeguards fail to keep pace with advances in the technology.
OpenAI has separately acknowledged that future models may reach higher levels of biological risk under its preparedness framework. The company said last year that upcoming systems could provide enough assistance to inexperienced users to warrant stronger protections and more rigorous testing.
Researchers involved in the MIT FutureTech study stressed that their findings are not predictions. Instead, they represent expert assessments intended to inform discussions around safeguards, testing and oversight.
