The BasePeople

Fabien Roger

Titles as the sources gave them. In 1 entry, with 3 checked quotations.

What they said

Newest first. Each quotation was found word for word in the source.

“When misalignment monitors fail, it is already often because of alignment failures, not because they are too weak. I expect this to continue in the future.”
Control monitors need alignment assessments tooLessWrong, October 24, 2026. Speaking as: Other. Opening summary of Roger's LessWrong post on monitors used in AI control.
“In the future, the most dangerous monitor failure modes will likely also be alignment issues.”
Control monitors need alignment assessments tooLessWrong, October 24, 2026. Speaking as: Other. Roger's prediction, followed by collusion between monitor and agent as one example.
“But getting evidence that monitors are aligned enough can still be hard, because the situations where the agent does sth bad are way more likely to be the ones where the monitor is also wrong.”
Control monitors need alignment assessments tooLessWrong, October 24, 2026. Speaking as: Other. Roger explains why monitor evaluations share difficulties with agent alignment evaluations.

In the Base

Every entry they appear in.

  1. EssayLessWrongWrote it

    Control monitors need alignment assessments too

    In a LessWrong post, Fabien Roger argues that AI systems used to monitor other AI agents often fail because of their own alignment problems, not weak capability, so monitors need alignment assessments much as agents do. The post lists possible research projects.