CAIN-42 E126 — Emerging General Intelligence

AGI NOT ESTABLISHED · overall PARTIAL · failures visible

Ten objectives, twenty-five gates, raw evidence, and an offline verifier. Novel-environment completion 87% vs 61% for the E125 engine; transfer +12.5 points with zero false accepts; collective zero Byzantine flips. Persistent learning failed its preregistered target. Active causal selection did not beat random. External review is BLOCKED.

FileContent
E126_REPORT.mdFull report
E126_SCORECARD.json25 gates
E126_NEGATIVE_RESULTS.mdFailures preserved
E126_LIMITATIONS.mdLimits
E126_REPRODUCE.mdCommands
E126_OFFLINE_VERIFIER.pyStdlib-only verifier
MANIFEST.jsonHashes