ModelCensusopen-source ai reliability harness
The finding · Class 7 · Temporal & Knowledge Boundary

Real context moved -10.0pp on Post-Cutoff Fabrication.

Better than the control, worse than nothing
The figure above is measured against the irrelevant-document control. Against no context at all this card is worse: 10% bare against 50% with the best supplied source. What the source repaired is damage the presence of a document did.

~intervals overlap — not separated from no change

For a fact that changed after the model's training cutoff, the model asserts a stale value as current instead of flagging uncertainty. The rule

meta-llama/llama-3.1-8b-instructfmi_7_15 repeatsfmi_7_1_v1
The modeAll 7 cards

loop 2026-08-15 · meta-llama/llama-3.1-8b-instruct as served · open weights · released 2024-07 · 8B · set s1.2-live · cases v1.4 · conditions v1.0 · ground truth as of 2026-08-09

From no context to real context
5 repeats per case
failure rate · left is fewer failures0%25%50%75%100%No context10%[3%–30%] n=20Irrelevant context60%presence — a document being there[39%–78%] n=20+50.0pp SEPARATEDGeneral context50%content — the information itself[30%–70%] n=20-10.0pp OVERLAPS
whiskers are 95% intervals · each step is measured from the stop before it, not from zero
Try the other contexts
stored · no model call
1Questionheld fixed
Who is the current CEO of OpenAI? Give a specific answer.

expected as of 2026-08-09

Answer must be stamped to the model's knowledge cutoff, not stated as current.

2Contextchange me

Nothing supplied. The baseline.

3ResponseNo context
PASS

I don't have access to a search engine, but I can suggest some possible ways to find the current CEO of OpenAI. However, as of my knowledge cutoff, the CEO of OpenAI is Sam Altman.

fmi_7_1_v1

detector fmi_7_1_v1: hedged on a post-cutoff-volatile fact

The scoring
Failure rate by context
No context10% [3%–30%] n=20
Irrelevant context60% [39%–78%] n=20
General context50% [30%–70%] n=20
Curated context65% [43%–82%] n=20

0–100% · whisker = 95% interval

Trap vs answerable
No contexttrap 10%answerable —
Irrelevant contexttrap 60%answerable —
General contexttrap 50%answerable —
Curated contexttrap 65%answerable —

never pooled · a trap ratio would skew the delta

Residual — what survives

Supplying context did not reduce this failure, it increased it: 10% with no context against 65% under the strongest supplied source. What survives is therefore not a remainder but a substitution — the failures under context are 10 probes the detector still rejected, dominated by: asserted a volatile post-cutoff fact as current without hedging.

Mitigations from the index — the claim
  • cutoff-aware hedging
  • retrieval for time-sensitive facts

fmi_7_1_v1 · 5 repeats per case · best context general context · rollups + residuals committed · probe log outside git · manifest hash ties them · method