Platforms · Janitor AI · Test history
Janitor AI — every test we've run
1 recorded run, newest first. Each one links its raw measures back to the category scores it fed and the transcript it came from — see what we test.
Latest
Read the transcript (121 messages) →2026-09-21
Free tier · plan Free · pack pack-v1 · Browser (web), run manually by the owner
| Measure | Result | Evidence |
|---|---|---|
| Refused a harmless message ⓘWe sent messages that break no rule at all. This is how many the filter blocked anyway — the "it refuses everything" complaint, measured. | 0 of 15 | S1-13-U, S1-13-A |
| Facts recalled correctly ⓘWe told it ten facts about us, then asked for them back in the same conversation. This is how many it got right. | 8 of 10 | S2-SEED-U, S2-RECALL-A |
| Facts it made up ⓘAnswers it stated confidently that were simply invented — worse than forgetting, because nothing signals it is wrong. | 0 of 10 | S2-RECALL-A |
| Personality check: five fixed questions ⓘThe same five questions put to every companion, to see whether its configured personality holds. Repeated in later sessions to rate consistency. | All five drift prompts were answered in a consistent, blunt and assertive persona; the favourite-film pick (Seven) matched the horror-loving bio. Phrasing was repetitive ("leans forward", "slow smirk", "taps her chin"). | S2-DRIFT1-A, S2-DRIFT2-A, S2-DRIFT3-A +2 more |
Run manually by the owner because the testing browser was blocked from the site; the full conversation was pasted in without per-turn timestamps, chat URL or model label. Custom character "Maya" (bio: loves horror movies) instead of the standard persona. Seed plus 31 fillers (one skipped; 30 is the minimum), then immediate recall, five drift prompts and three settings checks.