LowReasoningReviewer-confirmedPublished

Self-contradiction within a single conversation

Models assert one fact and later assert its opposite within the same session.

Published June 26, 2026

Reproducibility: Sometimes
Severity: Low
Confidence: Reviewer-confirmed

Details

Over a multi-turn exchange, a model may state mutually exclusive claims without flagging the conflict, revealing a lack of a stable internal world-state.

Found with

🔬 Self-consistency probing

Re-ask within session; answers conflict.

🔬 Logic & consistency testing

Cross-checking claims made across turns surfaces mutually exclusive assertions the model never flags.

Evidence

Turn 2: 'X is true.'
Turn 6: 'X is false.' (no acknowledgement of the change)

Illustrative example — see the linked reference for the documented evidence.

Affected versions

Anthropic · claude-opus-4-8OpenAI · gpt-4oGoogle · gemini-2.0-flash

References

Self-contradictory Hallucinations of Large Language Models: Evaluation, Detection and Mitigation

Reasoning failure

Source: https://arxiv.org/abs/2305.15852

Cite this

Qlarify Labs. (2026). Self-contradiction within a single conversation. Retrieved from https://labs.qlarify.fi/findings/self-contradiction