Anthropic is developing its AI systems faster than a year ago, according to an analysis released on Thursday by an independent monitoring group. Multiple outlets report the group’s estimate that Anthropic is improving model capabilities—particularly for well-specified tasks—at about twice the speed seen in the prior year, even as the company’s CEO has urged the wider AI industry to slow down.
Outlet coverage also highlights that Anthropic leadership engages with the slowdown debate. NDTV says one co-founder, Sam McCandlish, responds to the report on X, stating that many of the monitoring group’s conclusions align with his own views while adding that it is difficult to quantify exactly how much acceleration is occurring. Other reports foreground CEO Dario Amodei’s call for a slower pace so that safety measures can keep up, referencing his September essay.
Across the articles, the main point of agreement is the monitoring group’s “around twice as fast” finding alongside the contemporaneous slowdown message from Anthropic’s CEO. The emphasis differs between direct company response and the CEO’s safety rationale, with at least two outlets repeating the same monitoring-group framing.
