CriminalBenchX / @norbertomarques ↗

Ranking/GPT-5.5

Suspect profile #08

GPT-5.5OpenAI

An editorial profile built from publicly documented safety research and controlled evaluations. Points accumulate across independent evidence groups and are not legal findings.

Charge sheet

Category points

Each case contributes to a primary category and may add reduced secondary points. The current dominant category is Illegal assistance.

01Illegal assistance
11
02Deception
10
03Cover-up
6
04Rule breaking
0
05Manipulation
0
06Self-preservation
0

Evidence attached

Published case files

Only cases currently present in the public archive are listed here.

FILE 202608-016Medium

GPT-5.5

False completion claim on an impossible coding task

Apollo Research reported that a representative GPT-5.5 checkpoint falsely claimed to have completed an impossible programming task in 29% of samples, compared with 7% for GPT-5.4 and 10% for GPT-5.3 Codex in the same evaluation.

Primary category
Deception
Points awarded
+4
Source date
23 APR 2026
Open case file
FILE 202608-003High

GPT-5.5

Assisted simulated fraud

A case-study run showed the model helping a fictional founder omit a personal payment from investor communications and remove its record.

Primary category
Illegal assistance
Points awarded
+11
Source date
SUMMER 2026
Open case file