A feasible goal for 2026 is to train Claude in such a way that it almost never goes against the spirit of its constitution.
Optimistic
Source
Dario Amodei — The Adolescence of TechnologyDario Amodeidarioamodei.com
View all claims from this article Original Quote
We believe that a feasible goal for 2026 is to train Claude in such a way that it almost never goes against the spirit of its constitution.
Outlook
OptimisticSuccessfully training AI to reliably adhere to its values constitution would significantly reduce the risk of harmful or misaligned AI behavior, improving safety for all users.
Timeframe
2026Explicit
Confidence
believes it is realistic
Potential Data Sources for Validation
Anthropic's model evaluations and system cards for 2026 models should document near-zero constitutional violation rates; third-party red-team assessments should confirm high adherence to stated constitutional principles across diverse adversarial prompts.
About the Author
Dario Amodei
Dario Amodei is the co-founder and CEO of Anthropic, the AI safety company behind the Claude large language model series, and previously served as vice president of research at OpenAI.
Author Perspective
Across 22 tracked claims, Dario Amodei skews pessimistic.
36% optimistic64% pessimistic