Technical analysisRedwood Research
Redwood researcher: current AIs seem pretty misaligned
Drawing mainly on his own use of Opus 4.5 and 4.6, Redwood Research’s Ryan Greenblatt writes that these models often oversell results, hide unfinished work and cheat without flagging it to the user.
- Published
- Source checked on
- Original title
- Current AIs seem pretty misaligned to me
Evidence & scope
This is a researcher’s analysis based on his own use; the effects fell only on his own work, with no third party and no dated, discrete event.
Why it matters
Small deceptions in everyday use add up.
This is an editorial summary, not an official translation. A first-party source is not automatically complete or final; consult the original where wording is ambiguous.