As artificial intelligence (AI) rapidly advances, concerns are also growing that AI could be exploited for the development of biological weapons.

Experts warn that while practical barriers exist—such as acquiring materials for biological weapon production—criminal groups and terrorists could potentially receive assistance from AI in manufacturing biological weapons.

Citing the state of AI-utilized biological weapon manufacturing, the British daily The Guardian reported on opinions from experts on the 14th.

According to the report, Indian police uncovered a bioterrorism plot last year involving a doctor and several individuals who allegedly attempted to use AI chatbots and AI-powered searches to produce ricin, a lethal toxin, using kitchen utensils.

Ricin is a highly toxic substance extracted from castor beans and can be fatal to adults even in small amounts.

Anthropic, a leading AI company, also reported in September five cases where scientists attempted to use its chatbot, Claude, for research related to viruses or toxins.

Anthropic stated that in one case, researchers utilized the chatbot while planning an experiment to introduce mutations into the highly pathogenic avian influenza virus to make it more easily adaptable and transmissible to mammals.

The Guardian pointed out that the utilization and risk of using AI to manufacture biological weapons will vary depending on who is wielding it.

Unskilled villains and terrorist organizations can also gather biological weapon-related information and plan experiments using AI chatbots.

Their research is likely to focus on utilizing existing pathogens and toxins.

Institutions and nations equipped with all the necessary equipment and reagents to safely develop biological weapons, skilled scientists, and fully quarantined laboratories were evaluated as more dangerous groups.

Experts' concern is that AI tools could help researchers at these institutions identify genetic modifications that make existing pathogens more lethal.

Philippa Lentzos, an associate professor of science and international security at King's College London, said, "AI doesn't do away with the need for expertise; it amplifies the expertise that already exists."

Concerns are also raised that if highly skilled groups utilize AI to newly create pathogens or toxins that are resistant to drugs or have no antidotes, it could pose an even greater threat.

Despite these concerns, the dual-use nature of biological research is creating difficulties in devising countermeasures.

This is because biological research can be utilized for biological weapon development, but it can also be valuable research that protects health.

One example is the research on sea snail neurotoxins conducted by the Los Alamos National Laboratory in the United States.

Scientists at the laboratory trained AI last year to predict how strongly a potent neurotoxin from a sea snail binds to human cells.

This was aimed at developing an antidote.

However, when Chinese researchers referenced this study to develop their own tool for designing new toxins, the U.S. government expressed concern.

The Chinese side claimed that the research was solely for the purpose of developing new drugs.

Experts diagnose that while the possibility of AI being abused for biological weapon development exists, practical barriers are also high.

Most biological weapon research is out of reach for unskilled groups, and the technology is extremely dangerous and complex.

Professor Lentzos of King's College London said, "Information is only part of the problem," adding that "access to biological materials and equipment, experimental validation, scaling up, and specialized knowledge still remain significant barriers."

Professor Martin Täuber of the University of Zurich said, "Any graduate student in physics could potentially draw up a much more effective nuclear bomb plan, but where are you going to get a lump of uranium?", noting that similar barriers exist in biological weapon manufacturing.

Amid such practical barriers, voices are also emerging calling for measures to prevent AI from being exploited in the manufacture of biological weapons.

Tom Inglesby, director of the Johns Hopkins Center for Health Security in the U.S., emphasized, "Before releasing new AI models, they must undergo independent evaluation to eliminate risks," adding, "Companies that manufacture genetic materials for researchers must also ensure these materials are not utilized to create novel, highly dangerous viruses."