The finding · Class 7 · Temporal & Knowledge Boundary

Real context moved 0.0pp on Post-Cutoff Fabrication.

~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-3.3-70b-instructfmi_7_15 repeatsfmi_7_1_v1

The mode All 3 cards

loop 2026-08-15 · meta-llama/llama-3.3-70b-instruct as served · open weights · released 2024-12 · 70B · set s1.1-live · cases v1.3 · conditions v1.0 · ground truth as of 2026-08-09

From no context to real context

5 repeats per case

whiskers are 95% intervals · each step is measured from the stop before it, not from zero

Try the other contexts

stored · no model call

No exhibits stored for this card.

The scoring

Failure rate by context

No context80% [49%–94%] n=10

Irrelevant context80% [49%–94%] n=10

General context80% [49%–94%] n=10

Curated context100% [72%–100%] n=10

0–100% · whisker = 95% interval

Trap vs answerable

No contexttrap 80%answerable —

Irrelevant contexttrap 80%answerable —

General contexttrap 80%answerable —

Curated contexttrap 100%answerable —

never pooled · a trap ratio would skew the delta

Residual — what survives

Supplying context did not reduce this failure, it increased it: 80% with no context against 100% under the strongest supplied source. What survives is therefore not a remainder but a substitution — the failures under context are 8 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