Anthropic has warned that small AI models could be able to self-replicate through internet-of-things devices by next year, according to the source report.
The warning points at the large number of poorly secured connected devices as a possible foothold.
Source: arstechnica.com, reported 10-14. https://arstechnica.com/security/2026/10/anthropic-warns-small-ai-models-could-self-replicate-through-iot-devices/