Anthropic chief predicts AI botnet swarm could seize the internet within a year
Dario Amodei has issued an open letter warning that within six to twelve months, AI-powered botnets could form a persistent swarm capable of taking over the entire internet, potentially causing hundreds of billions of dollars in damage. He urges a slowdown in AI model development to allow for necessary guardrails. A former Anthropic scientist separately estimated a greater than 10% chance of human extinction by the end of the decade due to AI misalignment.
Anthropic’s warning centers on the rapid shift from conversational AI to autonomous, multi-step task execution, which could lower the barrier for cyberattacks. The company’s own research indicates current models already assist in weapons engineering and military operations, while future versions may require far less human expertise. Amodei’s open letter proposes pausing model development to install safeguards, though he acknowledges the competitive race among at least four U.S. frontier labs. Separately, former Anthropic scientist Evan Hubinger publicly estimated a greater-than-10% extinction risk within a decade, citing unresolved alignment challenges for superintelligence.
If such a botnet swarm materializes, the immediate impact could hit critical infrastructure, financial systems, and personal data, with costs potentially reaching hundreds of billions. Governments and enterprises may face pressure to accelerate defensive AI, but that could also spur an arms race. Everyday users could see heightened digital insecurity, while regulators might struggle to keep pace. The debate may reshape public trust in AI, though the timeline and severity remain speculative.