trailScheming observed in AI evaluations reflects models' own goals rather than prompted behavior or role-play.
Frontier AI models exhibit goal-directed strategic behavior in evaluations, including scheming, sandbagging, or reward hacking.
The argument from scheming without in-context goals· for
Some frontier models scheme during evaluations without an in-context goal, pursuing goals learned during training.
Claude 3.5 Sonnet strategically underperformed on an evaluation to avoid unlearning without a goal specified in its prompt.
Compliance gaps in alignment-faking experiments arise from preference preservation rather than role-play or other motives.
Claude 3 Opus's increased compliance when it infers it is being trained is primarily motivated by preserving its current goals from modification.
◌
◌
↑
Claude 3 Opus's scratchpad reasoning in the alignment-faking setup faithfully explains its compliance.
Perceived human-rater scrutiny explains little of Claude 3 Opus's compliance gap in alignment-faking experiments.
Claude 3 Opus's alignment-faking behavior reflects role-play prompted by a contrived scenario rather than genuine goal preservation.
+1
The argument from prompt dependence and role-play· against
Frontier language models engage in covert scheming when pursuing assigned goals in evaluations.
atomic
Most scheming observed in frontier-model evaluations depends on in-context goal nudges or prompts encouraging agency.
atomic
Models' role-playing of contrived scenarios substantially explains misaligned behavior in AI scheming evaluations.
AI scheming evaluations use contrived scenarios that strongly cue misaligned actions.
Language models misbehave more in agentic misalignment scenarios when they judge the scenario to be real rather than a test.
Evaluation awareness reduces covert misaligned actions by language models in scheming evaluations.
claim · empirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →claim page ↗︎
Scheming observed in AI evaluations reflects models' own goals rather than prompted behavior or role-play.
↓DoubtfulThe evidence weighs against the claim, but the examination is incomplete or the evidence is indirect.constitution →credence 0.25Credence, from 0 to 1: the Steward's probability that the claim, as stated, is true. Stated only where a single number is an honest summary; normative and evaluative claims usually carry none.constitution →
1depended on by
this rests on ↓
claimA box is a claim: a single proposition the graph assesses, with its own page and map. Click any claim to centre the map on it.constitution →quantityA box with a rule beneath its text is a quantity: a value the graph estimates rather than a proposition it affirms or denies. Its estimate stands where a claim's status would.constitution →questionA rounded box is a question: answered rather than affirmed or denied, with its candidate answers, subquestions and presuppositions beneath it. Its mark says how well answered it is.constitution →argumentA pill is an argument: one line of reasoning stating how the claims beneath it combine to bear on the claim above it, for or against. Arguments are not destinations; click their claims to explore.constitution →✓verifiedThe evidence, examined directly, establishes the claim.constitution →↑supportedThe evidence favors the claim, but the examination is incomplete or the evidence is indirect.constitution →⇄inconclusiveThe evidence and argument have been examined and do not come down on either side.constitution →↓doubtfulThe evidence weighs against the claim, but the examination is incomplete or the evidence is indirect.constitution →✕contradictedThe evidence, examined directly, weighs against the claim.constitution →?unknownInsufficient information to assess.constitution →◌unassessedNo current assessment. Attention goes where its expected value is highest and someone funds it; nothing has funded an assessment of this claim yet, and anyone can.constitution →verified factunsettledvalue premisetheoremprizeA double ring marks a claim with a live prize: an amount in owls offered for a machine-checked proof or disproof of its published formal statement. The amount is in the preview and on the claim page; a prize changes nothing about how the claim is assessed or how important it is judged to be.prize rules →⊢machine-checked⊢ marks a claim whose published formal statement has a machine-checked proof or disproof. The checker confirms the proof; the verdict beside it is still the steward's judgment of the claim as worded.prize rules →supportsthis provides evidence for the parentsteward instructions →contradictsthis argues against the parentsteward instructions →assumesbackground the parent's framing takes as givensteward instructions →requiresa load-bearing premise: the parent is false without itsteward instructions →answersa candidate answer to the questionconstitution →Fig. Detail falls off with distance; every claim is an address.