Anthropic Warns of Bids to Use AI for Biological Weapons Development

Anthropic Reports Rise in Misuse of AI Models for Biological Research

A US technology company, Anthropic, has released a report detailing instances of its artificial intelligence models being misused for research that could aid in the development of biological weapons. The report, issued on Thursday, outlines eight cases of misuse over the past eight months, including attempts to use the chatbot Claude for research related to vaccine development and biological pathogens.

According to Anthropic, the company was unable to determine whether the research was legitimate or malicious, as inquiry that leads to vaccines can also aid in engineering dangerous pathogens. In response, Anthropic has taken steps to limit access to its models, including banning accounts involved in suspicious activities.

The report highlights concerns over the potential dangers posed by AI development. A scientist attempted to use Anthropic’s models to engineer mutations to chikungunya, a mosquito-borne virus similar to malaria, with the intention of making it more harmful in live animals. The company was unable to determine whether this research was intended for military or civilian purposes.

Experts have called for stricter access controls to AI models, citing the need for trusted researchers to utilize these technologies. Anthropic’s head of threat intelligence, Jacob Klein, stated that the company is taking a cautious approach due to the uncertainty surrounding the intentions of some users.

The report also notes that evaluations of older models found they were not capable of assisting in dangerous biological research, while newer models can complete complex scientific work. The incident has sparked renewed debate over the responsible development and use of AI technology.

Source: Al Jazeera