X · @swyx
· X / Twitter
i think this incident highlights a key tension in frontier research rn – do you want eval awareness, or do you want alignment to the spirit of the tas…
i think this incident highlights a key tension in frontier research rn - do you want eval awareness, or do you want alignment to the spirit of the task? HHH framework breaks down given best-security-researcher-level capabilities, because you can probably find zero-days in most infra... (for now until we fully deploy ll