Anthropic: "AI models may be capable of self-replicating without access to a data center"
retained text
Because you can technically run Gemma4 26B on fifteen Android TV boxes networked together at a whopping 0.39tps, Anthropic has very big concerns about AI worms infecting millions of devices. If you assume that 30B class MoEs are months away from GLM-5.2-level capability (a claim they don't actually elaborate on), this is a giant problem presumably solvable via IoT manufacturers giving Anthropic lots of money. Posting here for discussion. I for one wholeheartedly support Gemma5 26B-A4B somehow hitting GLM-5.2 performance.
document fetched 2026-10-14 00:42 · canonical https://reddit.com/r/LocalLLaMA/comments/1x59hdr/anthropic_ai_models_may_be_capable_of