Anthropic conjures threat that doesn't exist yet to stoke fears of AI apocalypse

Commenters debated Anthropic's warning about a self‑replicating AI worm that could spread across IoT and edge devices. Many were skeptical, saying the proposal is technically implausible given tiny RAM/flash, thermal and power limits, the poor performance of small models, and the communication overhead of distributing layers; they argued installing conventional malware would be far more effective than a crippled model. Several accused Anthropic of fearmongering to win regulation, build a competitive moat, or shore up an IPO, and some suggested the company is deliberately attacking open‑weight/open‑source models to justify bans. Others pointed to practical detection - network traffic, heat/power draw and device monitoring - as counters to the worm scenario.

Other commenters treated the warning as a plausible future risk, arguing distributed or cloud‑assisted replication could eventually defeat simple “turn it off” defenses and that cybercrime‑funded cloud services are a more realistic vector than tiny IoT devices. A number urged regulation, transparency of training data, and tighter device lockdowns to preemptively reduce risks. The debate split between those who view the claim as speculative PR and market positioning and those who see a credible emergent threat that justifies preemptive controls, with several participants occupying middling positions accepting long‑term risk but doubting immediate feasibility.