Provenance record

Can scientists reliably tell if a novel is written using AI?

Nature (tier 1, academic) 2026-10-27T00:00:00.000Z Original ↗

Discourse valence

-33

Adverse

confidence 41% · 5 items · range -100 to +5

-100

-50

+50

+100

-33

Adverse readingFavourable reading

Consensus of 5 models from different labs. Spread 105 points, agreement low.

Existential and catastrophic risk Synthetic media and manipulation Concentration of power

Excerpt as ingested

Nature, Published online: 27 October 2026; doi:10.1038/d41586-026-03394-4 Computer scientist Tuhin Chakrabarty talks about detecting AI-generated prose — and why it matters.

Every model that read this

Model	Provider	Stage	Score	Conf.	Latency	Prompt	When
Llama 3.3 70B	Meta	analysis	0	20%	2592ms	v1.0.0 / m1.0.1	2026-10-27 18:17
GPT-4.1 mini	OpenAI	consensus	0	70%	2325ms	v1.0.0 / m1.0.1	2026-10-27 18:18
Claude Sonnet 5	Anthropic	consensus	-70	35%	4147ms	v1.0.0 / m1.0.1	2026-10-27 18:18
Gemini 2.5 Flash	Google	consensus	-100	0%	2312ms	v1.0.0 / m1.0.1	2026-10-27 18:18
Mistral Small 3.1 24B	Mistral AI	consensus	+5	80%	3460ms	v1.0.0 / m1.0.1	2026-10-27 18:18

Llama 3.3 70B · reading

No substantive claim about AI's effect on society.

evidence: anecdotal horizon: n/a

GPT-4.1 mini · reading

The text only indicates a discussion about detecting AI-generated prose and its importance but does not specify any social implications or effects on human flourishing. The implications depend on how detection might be used, which is not detailed here.

evidence: reported horizon: near capability 30 societal 0 existential 0 economic 0

Claude Sonnet 5 · reading

This is a brief interview teaser about ongoing research into AI text detection, with no specific findings or data presented. The implication is mildly concerning (authenticity challenges) but the text itself offers minimal substantive evidence beyond framing a question.

evidence: speculative horizon: n/a

Gemini 2.5 Flash · reading

evidence: reported horizon: n/a

Mistral Small 3.1 24B · reading

The article discusses the challenges of detecting AI-generated prose. It does not make a strong claim about the societal impact of AI.

evidence: speculative horizon: n/a

The chain

SOURCE     Nature (tier 1)
   ↓
DOCUMENT   dea77d02-2d40-4588-a9cd-32e99baa7e63
           https://nature.com/articles/d41586-026-03394-4
   ↓
EVIDENCE   0 extracted excerpts
   ↓
MODEL RUN  5 runs, methodology 1.0.1
   ↓
SCORE      -33  (Adverse)
   ↓
CONFIDENCE 41%