AI cannot independently create biological threats or deadly viruses, virologist says

MOSCOW, September 30. /TASS/. Artificial intelligence (AI) will not defeat humanity by creating a deadly virus against it because it cannot experimentally verify its theories and calculations, while biological predictions are not always borne out, Nikolay Nikitin, Doctor of Biological Sciences and professor at the Lomonosov Moscow State University Department of Virology, told TASS.

He noted that AI is already helping scientists identify promising compounds for drugs and develop new materials, serving as an auxiliary tool.

"If we are talking about using artificial intelligence to create biological threats and engineer deadly viruses, AI will not defeat humanity. In biology, one can perform as many calculations and make as many predictions as one likes, but until the effect of a virus on a specific cell culture has been studied in a laboratory, this data has no practical value. AI calculations in biology are not always accurate, and AI itself cannot conduct experiments. Therefore, neural networks will not surpass us in this area and will not independently create any biological threats," Nikitin told a TASS correspondent.

In September, US company Anthropic, which developed the Claude chatbot, reported detecting cases in which its AI models were used in potentially dangerous research that could be linked to the development of biological weapons. The company’s report said its safety systems had identified five such cases. In particular, Anthropic obtained data indicating that scientists had attempted to use the neural network to plan experiments to enhance the function of the chikungunya virus, adapt avian flu to mammals, and prepare a grant application on overcoming immune defenses against viruses of the orthopoxvirus genus, which cause smallpox.

After detecting the threats, the users’ accounts were blocked. The company decided not to disclose the countries the researchers were from or the scientific institutions they were affiliated with. The report emphasized that the company "does not claim that [the researchers] intended to cause harm."