Anthropic Blocks AI Misuse That Could Support Biological Weapons
Anthropic reported blocking malicious attempts to use its artificial intelligence models for activities such as cyberattacks and biological weapons research. The company emphasized that increasingly powerful AI systems are enabling even individuals without advanced technical skills to pose significant threats, prompting the addition of stronger safeguards in newer models. The report highlights specific cases where Anthropic's systems prevented misuse, including an attempt to use its AI to support gain-of-function research on the chikungunya virus. The company urged governments and competitors to collaborate in preventing such abuse as it prepares for a fall IPO. A researcher recently resigned over concerns about irresponsible AI development. Anthropic acknowledged that while older models were less capable of aiding dangerous biological research, newer systems present greater risks due to their advanced capabilities.
Topics: Artificial Intelligence Biological Weapons Research Cybersecurity Threats Ethics in AI