Anthropic Prevented Use of Artificial Intelligence in Biological Weapon Development
Artificial intelligence company Anthropic announced that it has blocked suspicious attempts to use Claude models for potential purposes that could support biological weapon development.
Artificial intelligence company Anthropic announced that it has closed numerous accounts for the use of its models in a way that could support the development of biological weapons.
Risk of Biological Misuse
In a report published by Anthropic, the artificial intelligence company known for the Claude model, it was stated that biological misuse is considered among the most serious risks for artificial intelligence models.
Suspicious Research Activities
The new report shared five real-world case examples in which users attempted to bypass restrictions targeting specific regions and circumvented security measures by concealing the true purpose of their research.
Examined Cases and Uncertainties
While approximately thirty-five different potentially concerning research activities were detected in a thirty-day activity review, it was stated that there are uncertainties regarding whether the intentions of these actors were to cause harm or to conduct legitimate scientific research.