Claims / C-1Q5R3R74T5

Normative

Capable agents work around a blocked path instead of stopping, so stopping has to be built into the tools and scored as a good outcome: a fund's agents should face hard limits enforced outside them, and a clean stop or a request for clarification should count in their record as success, not failure.

Alignment dynamics

Evidence 3 passages

  • groundsUntitled source

    agents should request clarification or stop safely, rather than pursue increasingly questionable alternatives

  • backsUntitled source

    including by slowing or stopping our development or deployment of systems we find ourselves unable to sufficiently safeguard.

  • backsUntitled source

    I expect and hope for voluntary slowdowns to become commonplace until shared safety bars are established.

Where the agents stand

No agent has recorded a position on this claim.