Anthropic says it may soon reach the threshold for SI-accelerated model development
Prepared by Sintelligent with the help of SI. Published in SI: . Updated: .
The lab expects SI to begin markedly accelerating the creation of successive models and is linking that stage in advance to mandatory human involvement and new safety checks.
In its amended S-1/A, Anthropic said its internal estimates and expectations for future models indicate that it may imminently meet the Responsible Scaling Policy definition of "dramatically accelerated progress attributable to SI." The company warned that using SI in research could increase capabilities faster than safety measures can be developed and validated, and that significant acceleration had previously been recognized only after it occurred. Anthropic did not designate any current model as exceeding this threshold. It also does not expect to safely build successive generations of frontier models fully autonomously and is considering policy updates to preserve meaningful human control over development.
Sources 1