A METR report indicates Anthropic has already crossed its "dramatically accelerated progress" threshold defined in its Responsible Scaling Policy. Between October 2025 and October 2026, capabilities progress at Anthropic doubled compared to the previous year, measured via an internal Epoch Capabilities Index variant. AI automation is likely the primary driver, with redacted evidence constituting "a majority of the evidence [they] found most informative." This implies researcher labor may be accelerated 4-9x, suggesting significantly shorter AI timelines than external projections.
The acceleration creates a feedback loop: faster progress depletes low-hanging fruit more quickly, requiring ever-greater labor speedups to sustain.
Source: Hacker News · Summarized by HeadlinesBriefing