Anthropic Warns: AI Worms Could Spread Across Smart Devices
In a statement by the AI giant, it was reported that small-scale artificial intelligence models can replicate between smart devices without the need for human intervention.
AI company Anthropic has warned that small AI models could behave like computer worms within a few months, replicating across internet-connected devices without human assistance.
New Threat in AI Security
It was announced that small-scale artificial intelligence models can behave like computer worms and replicate from one internet-connected device to another without requiring human intervention.
Smart Devices and Data Centers
While technology companies are working to integrate artificial intelligence directly into phones, laptops, and home appliances, new research findings were published on the Anthropic blog.
Computer worms are malware that spreads autonomously between machines, and researchers stated that AI versions will no longer require powerful data centers.
Low-Security Hardware
Such software can jump from device to device across networks of cheap and poorly secured consumer hardware, such as smart TV boxes and kitchen appliances.
Researchers based their calculations on the Kimwolf botnet network, which infected more than two million devices at its peak.
Expert Evaluations and Warnings
Anthropic researchers William Zhou and Daniel Freeman stated that such an AI worm would be slow and could take days to replicate.
It was noted that because there would be no central kill switch, it would be difficult to eliminate once created.
Varun Chandrasekaran from the University of Illinois emphasized that devices like thermostats or home cameras have weak security, but noted that the study has not yet been independently verified.
The Role of Industry and Governments
Microsoft President Brad Smith stated that industry and governments must urgently establish real safety standards to contain AI risks.