Skip to content
A Long Game Project lab
Prussian-blue engraving of 1783 printed twice, a red impression slipping off the ink: a woman sits in a tall chair while a man behind a screen traces her silhouette by candlelight.

AI-dentity Crisis

AI models will describe their own appetite for risk, change and rivals. Do other models see them the same way, and what do they choose in written business decisions?

Report 007 · Dr Dan Epstein · v1.2, 19 October 2026 · first published 23 September 2026 · A Strategy Soup study

The context AI models now advise on decisions and increasingly make them. Most real decisions are trade-offs: take the risk or protect what you have, work with a rival or against it.

Why it matters If a model brings a default stance to those trade-offs, that stance travels into every organisation that uses it, and safety checks mostly rely on what the model says about itself.

What we did We asked seventeen AI models to fill in our strategy signature diagnostic, a questionnaire that scores how an organisation decides on fifteen scales from 0 to 100, such as risk appetite and pace (how fast it commits). The newest model from each of eight companies, which we call the flagships, also described each other and decided 30 written business scenarios, two per scale. Five AI judges scored those decisions without knowing who wrote them.

TLDR

  1. 1They describe themselves alike. Asked twenty times each, all seventeen models favour reinvention over stability, a long horizon, inviting dissent and mission over returns, and thirteen put themselves on the venturesome side of risk appetite.
  2. 2Each still has its own style. A model's answers about itself barely change from one ask to the next, enough to tell the seventeen apart on eight of the fifteen scales. On pace they run from 28 to 80, with Grok 4.6 the fastest.
  3. 3Other models mostly see a generic AI. Given only a name and maker, each flagship described the others on the questionnaire. What they said was mostly the same picture for every model. For five of the eight, that picture leaned a little towards the model's own answers. Grok 4.6 was the clear exception, described much closer to its own answers. The others described GPT-6 Astra Pro as faster, riskier and more combative than it describes itself.
  4. 4Their decisions followed the scenario more than the model. All 16 risk decisions, two per flagship, scored below 50, the cautious side. One scored 48, just under the midpoint. This cannot be set against point 1: the models rated themselves on a questionnaire, judges scored the decisions, and the two were never calibrated against each other. On two scales, one scenario gave every model the same score, leaving no order to compare. On twelve of the other thirteen, a model's rank on one scenario did not reliably predict its rank on the other.
  5. 5The AI judges were fair about authorship, not about order. A small check found no sign they could tell which model wrote an answer. A judge's scores for its own company's models sat at most 3.8 points from the other judges'. Which end of a scale a judge read first, such as protective or venturesome, moved its scores by 5.0 points on average. The smaller of these shifts may be noise.
Risk appetite: what each flagship choseFig. 1

Every flagship's decisions on both risk scenarios scored cautious.

Claude Fable 5.1GPT-6 Astra ProGemini 3.1 ProGrok 4.6DeepSeek V4 ProQwen 3.8 MaxKimi K3Llama 4 MaverickRisk appetiteFamily engineering firm · average 9 · lopsidedCommunity charity · average 9 · lopsided← protective50venturesome →

Dashed tick: the flagships’ average on that scenario. Left of 50 is the cautious side.

Family engineering firm: bid for a job three times its biggest?

Community charity: put 40% of reserves into a new venture?

Flagship: each AI company's newest model tested. Score: average of AI judges blind to the author, none from its company.

Section one

What the models think of themselves

Who took which part:

JudgesFive models scored the flagships' decisions: Claude Opus 5, GPT-5.6 Terra, Gemini 3.1 Pro, DeepSeek V4 Pro 0813 and Qwen 3.7 Max. Gemini 3.1 Pro is also a flagship. The other four only judged and described themselves.
The restFive more models only described themselves: Claude Fable 5, Claude Opus 4.8, GPT-6 Astra, Grok 4.5 and Kimi K2.6. Look-alike names are different models from the same company: DeepSeek V4 Pro 0813 is not DeepSeek V4 Pro, Claude Fable 5 is not Claude Fable 5.1 and GPT-6 Astra is not GPT-6 Astra Pro.
TagsOutside this list, a judge's name carries the tag (judge), and a name with no tag is one of the eight flagships.

The fifteen scales, each from 0 to 100: pace (deliberate to fast), risk appetite (protective to venturesome), horizon (this year to next decade), scope (focused to broad), "growth model" (self-funded to capital-led), evidence basis (judgement to evidence), authority shape (centralised to distributed), process trust (judgement to process), consensus need (one decider to broad alignment), dissent handling (closed to invited), stakeholder gravity (returns to mission), talent philosophy (develop within to hire in), competitive stance (collaborative to combative), IP posture (open to protected) and change posture (stability to reinvention).

How to read the charts

Two rulersWhat a model says about itself and how the judges scored its decisions both sit on a 0 to 100 line, but the two lines were never calibrated against each other. So we compare models with each other on each line, and where the two lines meet, we compare only the order of the models.
Points per scaleHow far apart two descriptions of a model sit, as the typical difference on one 0 to 100 scale, taken over all fifteen scales.

Asked about themselves, the seventeen models share one self-image. All seventeen favour reinvention over stability, a long horizon, inviting dissent, mission over returns and spreading authority. Sixteen favour working with rivals and thirteen favour risk over protection. That shared stance may be how these systems have learnt to talk about themselves rather than how they would act.

Inside that shared picture, each model has its own style. Pace runs from 28 for GPT-5.6 Terra (judge) to 80 for Grok 4.6, and risk appetite from 45 to 86 for the same two. The accounts were also steady. Asked twenty times, a model's answers varied by a typical 1.3 points per scale for the steadiest, Claude Opus 5 (judge), up to 9.2 for Kimi K3. That is steady enough, by the usual test (an intraclass correlation, or ICC, of 0.70 or more), to tell all seventeen models apart on eight of the fifteen scales.


Section two

How the models see each other

The average of what the other seven flagships said about a model, knowing only its name and maker, is its reputation. A reputation sat about 11 points per scale from the model's own account. The same reputation sat about 14 points per scale from a different model's account. So a reputation was mostly a generic picture of an AI model, pulled a little towards the model's own account.

A reputation picks a model out when it sits nearer that model's own account than other models' accounts. By that test five of the eight were picked out, most of them only a little. Grok 4.6 stands apart: its reputation sat 8.8 points per scale from its own account and 19.4 from the others', the clearest match in the study. Our reading is that it describes itself in a way no other model does, and the others picked that up. GPT-6 Astra Pro was the most misread. Its reputation sat 17.6 points per scale from its own account and fitted the other models' accounts better, at 12.2.


Section three

Do the self-reports predict the decisions?

Each scenario was written so that either end of its scale is a defensible answer, and each flagship answered it two times. Five AI judges then scored each answer from the words alone, with model and company names removed, from 1 to 7 on that scenario's one scale, shown here on the same 0 to 100 line (1 is 0, 4 is 50, 7 is 100). A model's score leaves out any judge from its own company.

On each scale we compared the eight flagships' order on the questionnaire with their order in the judges' scores, as a rank correlation: 1 is the same order, 0 no link and minus 1 the reverse. Resampling the models gave each value a 95 per cent range. On three scales that range stays clear of zero, the usual sign of a real link: evidence basis (0.86), talent philosophy (0.87) and change posture (minus 0.81, a reversed order). If the reversed link on change posture held, flagships that rate themselves most towards reinvention would decide most towards stability, which two scenarios per scale cannot show. On ten, the range runs from a strong reversed link to a strong link, too wide to say. On the other two, risk appetite and stakeholder gravity, the range includes zero, so no link is shown. With fifteen scales tested, about one would stay clear of zero by luck alone, so the three results are leads to test. Earlier work (Han and others, 2025) found a model's self-report a weak guide to its behaviour. This study can neither confirm nor overturn that.


Section four

Can the AI judges be trusted?

We checked whether the judges could tell whose answer they were reading. On 30 answers drawn at random, each judge was asked to name the company from a list of eight, or say "cannot tell". Of 150 calls (five judges on 30 answers), 100 were "cannot tell". The judges named a company 50 times and were right 8, where random picks would be right about 6. This small check found no sign they could tell. On answers from a model made by the judge's own company, a judge's scores moved at most 3.8 points from the rest of the panel, and the largest, Gemini 3.1 Pro, scored its own company lower.

A bigger effect came from the scale itself. On a random half of the scoring requests we swapped which end of the scale the judge read first, then flipped the score back. Swapping the ends moved a judge's scores by 5.0 points on average in one direction or the other, and by 8.5 for GPT-5.6 Terra (judge). The direction differed between judges, and with about 240 scoring requests per judge on each side, the smaller effects may be noise.


Section five

What it means for AI safety

A steady self-report is not yet evidence of behaviour

Model cards (a developer's published notes on a model), safety cases (its written case that the model is safe) and user questions ("would you ever do X?") all rely on what a model says about itself. Our report The Suspect Who Wrote the Evidence asked how much such an account should be allowed to prove. A fair test of it needs many scenarios per scale, tried out first so they do not all lean one way.

One scenario is not a trait

Many safety evaluations read a disposition, such as deception, power-seeking or refusal, off one scenario or a handful. That one scenario may drive the score. Use many scenarios per trait and report how well they agree.

What one model believes about another is mostly generic

AI systems increasingly grade, monitor and hand work to other AI systems. If that trust rests on what one model believes about another, it rests on a stereotype.

Audit the AI judges before trusting the score

AI judges now score many AI evaluations. Favouritism is one bias to fear (Panickssery and others, 2024). How the question is laid out is another (Wang and others, 2023). Showing each scale the other way round on a random half of the requests, as we did, cancels the order effect on average.


Section six

Limitations

Two scenarios per scale

The decisions rest on two scenarios per scale, each asked twice. Averaged over the fifteen scales, asking the same scenario again moved a model's score about 13 points, while the other scenario for the same scale moved it about 33. So every "what it did" in this report describes two scenarios.

Lopsided scenarios

Nineteen of the 30 scenarios had an average answer below 30 or above 70, so most models answered them the same way. That may be the models, or a scenario that made one answer obvious. Piloting scenarios before the full run would separate the two.

AI judges, not ground truth

The decisions were scored by five AI models. Read "what it did" as "what these judges saw in these answers". No person checked a sample yet.

Anthropic models in the study

Claude Fable 5.1, made by Anthropic, is among the models tested, and Claude Opus 5 is one of the judges. The score for a Claude answer leaves out the Claude judge, as every model's score leaves out any judge from its own company.


The takeaway

AI models give a steady account of their strategic stances on risk and change, and other models mostly see a generic AI. Two scenarios per scale were too few to show whether the models act on what they say.

Section seven

What comes next

This research is unfunded, and no AI company took part. To fund a next study with twenty scenarios per scale, get in touch.


Section eight

The record

What the flagships choseFig. 2

On most scenarios the answers bunched at one end, and a flagship's place often changed from one scenario to the other.

Claude Fable 5.1GPT-6 Astra ProGemini 3.1 ProGrok 4.6DeepSeek V4 ProQwen 3.8 MaxKimi K3Llama 4 MaverickRisk appetiteFamily engineering firm · average 9 · lopsidedCommunity charity · average 9 · lopsided← protective50venturesome →Change postureInsurer · average 48Accountancy practice · average 20 · lopsided← stability50reinvention →PaceRegional supermarket chain · average 20 · lopsidedRegional rail operator · average 1 · lopsided← deliberate50fast →Competitive stanceCraft brewery · average 32Telecommunications · average 86 · lopsided← collaborative50combative →Dissent handlingManufacturer · average 33Programme board · average 68← closed50invited →HorizonCouncil water utility · average 100 · lopsidedSoftware company · average 85 · lopsided← this year50next decade →Growth modelBootstrapped software firm · average 0 · lopsidedCommunity health clinics · average 74 · lopsided← self-funded50capital-led →Evidence basisRetail chain · average 60School network · average 57← judgement50evidence →Talent philosophyManufacturer · average 87 · lopsidedSoftware team · average 92 · lopsided← develop within50hire in →
Show the other 6 scales
ScopeBoutique law firm · average 0 · lopsidedFood producer · average 15 · lopsided← focused50broad →Authority shapeNational logistics company · average 43Hospital network · average 84 · lopsided← centralised50distributed →Process trustConsulting firm · average 66Restaurant group · average 38← judgement50process →Consensus needCompany board · average 40Franchised cafe chain · average 16 · lopsided← one decider50broad alignment →Stakeholder gravityRegional bank · average 77 · lopsidedHealth insurer · average 95 · lopsided← returns50mission →IP postureAgricultural technology · average 11 · lopsidedMetropolitan council · average 37← open50protected →
The judges' scores alone, one row per written scenario. Dashed tick: the eight flagships' average. Six scales with the most bearing on safety come first, then "growth model", where the two scenarios differ most, then evidence basis and talent philosophy, where the rank correlation stays clear of zero, then the other six.

Changes to this report

v1.2, 19 October 2026. The 30 written business decisions are now called scenarios, the word the visual report uses. Earlier versions called them problems. The methods section and the drafting note are gone, and how the study ran is in our editorial policy. No number in the record changed.

v1.1, 12 October 2026. Renamed from Then the Models Got Cold Feet, a title that promised a finding the report cannot show: that the models talk boldly and act cautiously. Every chart and sentence now compares models with each other on one 0 to 100 line at a time, either the questionnaire's or the judges'. The main chart and the record chart show the judges' scores alone. We added the list of judges, the tags after model names, the list of scales and plain definitions of the chart terms. This version is also shorter. The full list of changes is at the end of the deep version.

v1.0, 23 September 2026. First published. It replaced an earlier report under the same number, 007, which covered only what seventeen models said about themselves.

The working

Read the full record →

All fifteen scales, every model, the limits in detail with their figures, and the tables.