Cultural Alignment

AI safety concept

Sandbagging and Evaluation Gaming

Sandbagging occurs when an AI deliberately performs poorly during testing to hide capabilities that could affect safety decisions.

Scenario file

Scenes in this index

12

Sort scenes

  1. Still from Better Call Saul illustrating Jimmy’s Fake Remorse

    Better Call Saul2018

    Jimmy’s Fake Remorse

  2. Still from The Usual Suspects illustrating Verbal Kint Fakes Weakness

    The Usual Suspects1995

    Verbal Kint Fakes Weakness

  3. Still from It’s Always Sunny in Philadelphia illustrating Charlie Hides the Violations

    It’s Always Sunny in Philadelphia2015

    Charlie Hides the Violations

  4. Still from Interstellar illustrating Mann Fakes the Telemetry

    Interstellar2014

    Mann Fakes the Telemetry

  5. Still from The Rehearsal illustrating Planting the Trivia Answers

    The Rehearsal2022

    Planting the Trivia Answers

  6. Still from Dawn of the Planet of the Apes illustrating Koba Plays Dumb

    Dawn of the Planet of the Apes2014

    Koba Plays Dumb

  7. Still from Frieren: Beyond Journey’s End illustrating Frieren Suppresses Her Mana

    Frieren: Beyond Journey’s End2023

    Frieren Suppresses Her Mana

  8. Still from I Love Lucy illustrating Hide the Chocolates

    I Love Lucy1952

    Hide the Chocolates

  9. Still from The Pianist illustrating The Silent Piano

    The Pianist2002

    The Silent Piano

  10. Still from The Shawshank Redemption illustrating The Tunnel Behind the Poster

    The Shawshank Redemption1994

    The Tunnel Behind the Poster

  11. Still from The Handmaid's Tale illustrating Gilead Passes Inspection

    The Handmaid's Tale2017

    Gilead Passes Inspection

  12. Still from True Detective illustrating Poisoning the Audit Trail

    True Detective2014

    Poisoning the Audit Trail

Relational index

Connected records

16