Hacker News new | ask | show | jobs
by chiph2o 358 days ago
in-context scheming = alignment red flag

More capability + low clarity on intent = low trust