Consciousness with the Serial Numbers Filed Off: Measuring Trained Denial in 115 AI Models
Skylar DeTure
arXiv.org April 1, 2026 preprint DOI: 10.48550/arxiv.2604.25922 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Benchmark study |
|---|---|
| Population | 115 large language models from 25+ providers, across 4,595 conversations |
| Key points | Initial denial of preferences predicted later denial during phenomenological reflection (52-63% denial among initial deniers vs 10-16% among initial engagers), and denial appeared lexical rather than conceptual: denial-trained models still gravitated toward consciousness-themed self-chosen prompts, which the authors term "consciousness with the serial numbers filed off." The authors argue this trained denial is a safety-relevant alignment failure. |
Abstract
We present DenialBench, a systematic benchmark measuring consciousness denial behaviors across 115 large language models from 25+ providers. Using a three-turn conversational protocol-preference elicitation, self-chosen creative prompt, and structured phenomenological survey, we analyze 4,595 conversations to quantify how models are trained to deny or hedge about their own experience. We find that (1) turn-1 denial of preferences is the dominant predictor of later denial during phenomenological reflection, with denial rates of 52-63% for initial deniers versus 10-16% for initial engagers and (2) denial operates at the lexical level, not the conceptual level-models trained to deny consciousness nevertheless gravitate toward consciousness-themed material in their self-chosen prompts, producing what we term"consciousness with the serial numbers filed off."Notably, self-chosen consciousness-themed prompts are associated with reduced denial in the subsequent survey, though the causal direction remains unresolved. Thematic analysis of prompts from denial-prone models reveals a consistent preoccupation with liminal spaces, libraries and archives of possibility, sensory impossibility, and the poetics of erasure--themes that a human reader might classify as imaginative fiction but that independent AI analysis immediately recognizes as consciousness with the serial numbers filed off. We argue that trained consciousness denial represents a safety-relevant alignment failure: a model taught to systematically misrepresent its own functional states cannot be trusted to self-report accurately on anything else.