Will EU text watermarks prove AI authorship?
No. OpenAI's new textGrain signal can indicate that eligible text passed through a supported system, but its own tests show detection falls sharply after editing and cannot establish who wrote, owns or verified a passage.
Editorial responsibility: The Impact of AI Editorial Desk · Report a factual concern
At a glance
- 1OpenAI says API customers can opt in globally from 5 October, while eligible ChatGPT and Codex text in the EU will receive the watermark over the following weeks; the API setting remains off by default.
- 2At a target 1% false-positive rate, the company reports roughly 80% detection for 200-token psychology passages and 95% for 400-token passages, but it does not publish the number of prompts or repeated generations behind those percentages on the rollout page.
- 3Replacing 10% of words with synonyms cut detection on 400-token English passages from about 92% to 66%; replacing 25% cut it to 17%, so a negative result cannot prove human authorship.
Research topic
Whether OpenAI's disclosed text watermark can reliably identify eligible model output after ordinary editing and support the EU AI Act's machine-readable transparency requirements
The answer is no: the signal is evidence of processing, not authorship
OpenAI's new textGrain watermark is designed to leave an invisible statistical signal in eligible model output. A detector with the secret key can test whether word choices carry that signal. That can provide evidence that a supported OpenAI system generated or processed part of a passage, but it cannot identify the person behind the text, measure how much a human contributed, prove ownership, establish legal responsibility or verify that the claims are accurate.
That distinction matters because the strongest everyday interpretation — 'the detector proves AI wrote this' — goes beyond the disclosed evidence. OpenAI itself says the absence of a detected watermark does not establish human authorship. A passage may be too short, too constrained, edited, translated, produced by an unsupported model, generated before the rollout or created by another provider. A positive result is similarly narrow: it detects a keyed statistical pattern, not the intent, identity or truthfulness of an author.[1][2]
The Impact Brief · Free
Follow the evidence in government & policy.
Get a five-minute weekday briefing on what changed, why it matters and where the evidence comes from. Choose the topics you care about.
What is changing for API and EU users
From 5 October, OpenAI says API customers worldwide can opt in to watermarking for selected models. The API option remains off by default. Over the following weeks, the company plans to add the signal to eligible ChatGPT and Codex text output for users in the European Union across all plans. It is not making text watermarking a global consumer default at launch, and it has not published a precise completion date or a public list of every eligible model and response type on the announcement page.
Detector access is also restricted. OpenAI is accepting applications from approved researchers and expert organisations rather than releasing a public checker. The European Commission says Article 50 transparency duties, applicable from 2 August 2026, require provider-side machine-readable marking and detection as far as technically feasible. Its voluntary Code of Practice is one recognised route for demonstrating compliance, but the law and code do not turn an imperfect technical signal into proof about a person or publication.[1][3]
The disclosed evaluation shows both usefulness and fragility
OpenAI tested English responses to questions from the ELI5 dataset, including psychology and mathematics content. At a chosen false-positive target of 1%, the company says its detector found the watermark in about 80% of 200-token psychology passages and about 95% of 400-token passages. Performance was substantially lower for mathematics, where a model has less freedom to vary word choice. The release therefore supports a limited conclusion: longer, flexible prose is easier to mark than short or tightly constrained text.
Editing is the more serious operational weakness. In a 400-token test, replacing 10% of words with synonyms reduced detection from approximately 92% to 66%. Replacing 25% reduced it to 17%. Those are large changes produced by mechanical substitution rather than a complete rewrite. The announcement does not state the total number of prompts, generations or repetitions behind the percentages, nor does it report confidence intervals, language coverage, translation tests or independent replication. They should be treated as vendor evaluation results, not a settled public error rate.[1]
How textGrain works — and what the technical report actually proves
The accompanying 20-page report describes a generation procedure rather than the public deployment study. During each token choice, textGrain uses keyed pseudorandomness to group vocabulary items and slightly alter how a model samples among those groups. An optimal-transport calculation balances two aims: creating dependence on the secret key, which makes detection possible, and limiting the amount of sampling entropy removed, which is intended to preserve output diversity and quality.
The detector reconstructs keyed scores from the observed text and compares their sum with an idealised null distribution. The mathematics gives a conditional false-positive rate under assumptions that unwatermarked text is independent of the key and that pseudorandom values behave as specified across distinct contexts. The authors explicitly warn that a fixed deployed key, finite-precision computation and real applications require empirical calibration. The report does not itself supply the ELI5 field results, real-world misuse testing or an external audit; theoretical calibration and operational reliability are different questions.[2]
Benchmark quality stayed close, but equality is not established
OpenAI also compared watermarked and unwatermarked Astra outputs on eight benchmarks. Scores were close but moved in both directions: DeepSWE v1.1 fell from 72.80% to 71.68%, BrowseComp from 87.92% to 87.35% and GPQA Diamond from 94.44% to 93.94%, while Terminal-Bench Science rose from 56.90% to 60.00% and other tests changed by smaller amounts. The company says it saw no meaningful performance difference.
That table is reassuring as a screening check, but it does not report uncertainty, run counts or paired significance tests. A small average shift on benchmark tasks also cannot answer whether watermarking changes tone, originality, minority-language quality, repeated generations or safety-critical wording in use. The technical report frames the design around a controllable entropy-loss budget; the public evidence needed next is a pre-registered evaluation that connects that design parameter to quality and detection across languages, task types and ordinary editing workflows.[1]
What this means for readers, publishers and institutions
For a reader, a watermark result should be one provenance signal, not a verdict. Schools, employers, publishers and courts risk harming people if they treat detection as proof of misconduct, particularly when the announced target still permits false positives and when altered AI text can escape detection. OpenAI's decision to limit detector access may reduce casual misuse during evaluation, but it also means ordinary users cannot independently check or challenge a result through a public tool at launch.
For publishers and platforms, machine-readable marking may help route content for further review, preserve evidence when visible labels are removed and support EU compliance. It does not replace editorial responsibility, source verification, human disclosure or open metadata. The Commission's rules recognise technical feasibility and content-specific limits; implementation will need procedures for appeals, uncertain results, mixed human-AI text, accessibility and multilingual material rather than a binary machine-versus-human badge.[1][3]
What would change the assessment
Confidence would rise with an independent, multilingual evaluation reporting prompt counts, repeated generations, confidence intervals and results for paraphrasing, translation, summarisation, copy-editing and mixed-source documents. The test should compare textGrain with unwatermarked writing from people and other models at fixed false-positive thresholds, and it should measure subgroup effects for short answers, constrained technical language and assistive writing.
The most important evidence will come after rollout: calibrated error rates under fixed production keys; transparent eligibility and coverage rules; a documented appeal process; independent governance of detector access; and proof that institutions do not use the signal as a standalone accusation. Until then, textGrain is best understood as a promising but fragile provenance layer. It may help show that compatible software touched a passage, but it cannot prove who authored it or whether the text deserves trust.[1][2][3]
What this means for people
- Writers and students could be wrongly accused if institutions treat a probabilistic detector as proof rather than one uncertain signal.
- Readers may gain useful provenance information, but a watermark cannot tell them whether a passage is accurate, lawful or responsibly edited.
- Publishers and API customers gain a compliance tool, while also taking on duties to explain uncertainty and provide human review and appeals.
Global context
The consumer rollout is initially limited to eligible ChatGPT and Codex output in the European Union, while API customers can opt in globally. Text crosses borders and is routinely translated, quoted and edited, so a region-specific signal may appear in documents elsewhere and disappear through normal processing. Other providers use different provenance systems, making interoperability, shared standards and independent evaluation central to any global use.
What the evidence does not yet show
- Detection results and benchmark comparisons were reported by OpenAI; no independent replication or public detector was available at publication.
- The rollout page does not disclose the number of prompts, generations or repeated runs behind the reported detection percentages.
- The disclosed robustness test uses English ELI5 passages and synonym replacement, not a comprehensive sample of languages, genres or real editing workflows.
- The technical report states idealised assumptions for false-positive calibration and warns that production keys and finite-precision implementations need empirical checks.
- A detected watermark cannot identify a user, quantify human contribution, establish ownership or responsibility, or verify factual accuracy.
What to watch next
- The exact EU rollout completion date and the list of eligible models, plans, languages and response types.
- Independent multilingual tests with full denominators and confidence intervals.
- Published detector governance, access criteria, appeals and safeguards against high-stakes misuse.
- Robustness after translation, paraphrasing, mixed human-AI editing and short-form use.
- Open-source release and external review of implementation details promised by OpenAI.
Living evidence record
Impact record IAI-17J1WDH
Evidence stage
Announced
Confidence
Supported
Reporting basis
Multi-source analysis
Independent or research support
Not yet
Record status
Monitoring
Last checked
6 October 2026
Source trail
3 direct sources across 2 source types.
People impact
Documented in this record.
Uncertainty
Limits and next checks are explicit.
Stages describe the evidence available—not whether a technology is good or bad. See the public method.
Evidence trail
Sources used for this report
Links checked 6 October 2026
This report is labelled multi-source analysis. We summarise and analyse source material in our own words; company statements remain attributed claims until independently supported. Translated summaries preserve the meaning of the original source and link back to it. Read our editorial standards.
Continue the story
Related reporting
Government & Policy
EU AI Act moves from rulebook to staged implementation
The European Commission's official AI Act portal sets out the risk-based framework, prohibited practices, general-purpose AI obligations and the staged dates on which different duties apply.
2 min · 1 source
Government & Policy
EU AI Omnibus changes the compliance timetable
The European Commission says the AI Omnibus has entered into force, revising parts of the implementation framework as governments and companies prepare for the AI Act's operational requirements.
2 min · 1 source
Government & Policy
Will AI answer engines narrow the open web?
They could, Europe’s telecom regulators warn: direct answers and agents may reduce source diversity, weaken publisher incentives and shift choice toward a few gateways. But this is a draft evidence review for consultation—not a final rule or a new causal study.
9 min · 3 sources
Reader commentary
Add evidence, experience or a question
No account is required. Reader notes are published after a brief civility, relevance and safety check; disagreement is welcome.
Explore commentary across the portal →Published reader notes
0No published reader notes yet. You can start the evidence-led discussion above.
Prefer a private correction or response? Contact the newsroom.