AIBleepingComputerabout 1 hour ago
Anthropic Warns of Self-Replicating AI Risk
3 min read
Anthropic warns that small AI models may self-replicate on consumer IoT devices by the end of 2026. Experiments using models like Qwen3.5-9B and Gemma4-26B-A4B showed success rates rising from 45% to 84% on Android TV boxes. The company estimates models with 30 billion parameters could reach this capability, urging stronger IoT security defaults to defend against potential distributed botnets.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyAnthropic
