AI company Anthropic says it has blocked multiple attempts to misuse its artificial intelligence systems, including activities involving cyberattacks, surveillance, propaganda and potentially dangerous biological research.

Advanced AI Misuse Raises New Security Concerns

In a new safety report, Anthropic said its latest investigations identified increasingly sophisticated efforts to use AI for harmful activities. The company highlighted cases involving cyber operations, surveillance and influence campaigns.

One particularly concerning case involved a request connected to research on the chikungunya virus. Anthropic said the request sought assistance with gain-of-function research that could potentially make the virus more transmissible or capable of evading immunity.

The company said newer AI models require stronger safeguards because their capabilities are significantly greater than those of earlier systems.

AI Safety Debate Intensifies in Washington

The disclosure comes as concerns about AI safety and regulation in the United States are increasing. Lawmakers and technology companies are debating how to prevent advanced AI systems from being exploited while allowing legitimate research and innovation to continue.

Anthropic said stronger safeguards and greater cooperation between technology companies and governments will be necessary as AI capabilities advance.

The latest report adds to growing pressure on the U.S. technology industry to demonstrate that increasingly powerful AI systems can be developed with effective security, misuse prevention and accountability measures.