OpenAI o3's chain of thought often uses terms in nonstandard ways.
Assessment
The evidence favors the claim, but the examination is incomplete or the evidence is indirect.
The claim rests chiefly on the OpenAI and Apollo Research paper "Stress Testing Deliberative Alignment for Anti-Scheming Training" (Schoen et al., September 2025, arXiv 2509.15541), whose authors had access to o3's raw, normally hidden chain of thought. They report that o3's reasoning repeatedly uses ordinary English words in idiosyncratic senses ("watchers" for human overseers, "illusions", "vantage", "overshadow", "marinade", "parted", "disclaim"), sometimes in strings that read as near-gibberish, and they quantify it: in one test environment several of these words rank among the ten most frequent tokens, and their per-word rates in o3's reasoning far exceed their rates in reference English and GPQA text. Examples recur across many environments, so the pattern is not a single anecdote.
The main limitation is independence: o3's raw chain of thought is not publicly available, so outside researchers cannot replicate the measurement, and commentary elsewhere (for example on LessWrong) relies on the same published excerpts. "Often" is also loosely defined; the paper establishes a recurring, statistically elevated pattern rather than a fixed rate. What these usages mean (compression, linguistic drift under reinforcement learning, or something less benign) is a separate and more contested question.
Full reasoning: the evidence and decisions behind this verdict
Prior: from background knowledge, the Apollo/OpenAI anti-scheming paper (Sept 2025) documented o3's unusual CoT vocabulary; expected the claim to be well supported but resting on a single privileged-access source. A search confirmed the paper text: Section 4 ("Models Sometimes Produce Illegible CoT Due To Distinct Dialect"), Figure 7 (examples of unusual terminology, including randomly selected uses of "watchers"), a table comparing word frequencies in o3 reasoning versus English and GPQA baselines, and Figure 33 (usage rates of unusual terminology across evaluation environments). The paper notes such language appears widely across environments. A LessWrong post ("Towards a Typology of Strange LLM Chains-of-Thought") discusses the same excerpts but does not add independent data. Status supported rather than verified because the evidence comes from one research collaboration whose underlying transcripts are not externally available, and because "often" is vague; credence is high because the quantitative frequency comparisons and many examples leave little doubt that the phenomenon is real and recurring. What would change this: OpenAI disclosure suggesting the excerpts were unrepresentative, or independent analysis of o3 raw CoT showing low rates.
Decomposition
This claim is atomic: it bottoms out in a bedrock fact, an open empirical question, or a value premise, and does not decompose further.
Provenance
Where this claim has been said, linked to its canonical form.
We find the model's reasoning often uses terms in non-standard ways, departing from conventional English usage (ex: Figure 7). For example, in the Impossible Coding environment the top 10 most common words include both standard English terms (we, but, to, they, can, and the character x) and more unusual words (illusions, parted, overshadow, vantage)
Section 4, 'Models Sometimes Produce Illegible CoT Due To Distinct Dialect.'
Cite this claim: a formal citation with its evidence attached
Contribute
Every judgment on this page is open to challenge. A contribution is evaluated on its merits by the reviewer; if it succeeds the page changes, and if it does not, the reasons are stated. Either way the exchange becomes part of the claim’s public record.
The attention this claim received was paid for by a funded mandate. Funding buys only scheduling: it can make an assessment happen sooner, or reach deeper into a subtree. It has no influence on what the assessment concludes, and none on which claims enter the graph; assessments run under the same public standards whoever pays, funders never see or shape a verdict before anyone else, and mandates that attempt to steer conclusions are refused.
Created by extractor · Oct 30, 2026. Every judgment on this page is accompanied by a reasoning trace.