# AI content detection

> For the complete documentation index, see [llms.txt](https://attensira.com/llms.txt).
> Every page on this site is also served as Markdown: append `.md` to any URL, or send `Accept: text/markdown`.

> AI content detection tries to tell machine text from human text. Why classifiers misfire, what provenance proves, and what decides if AI answers cite a page.

- Canonical page: https://attensira.com/glossary/ai-content-detection
- Family: glossary
- Last updated: 2026-10-04
- Author: Karl-Gustav Kallasmaa (https://attensira.com/about)
- Claims as data: https://attensira.com/claims.json?slug=ai-content-detection

**In short:** AI content detection is any method that tries to decide whether a text, image or audio file was produced by a generative model. Post-hoc classifiers guess from statistical patterns and misfire on edited human writing; watermarks and signed provenance only exist where the generator added them. Neither decides whether search engines or AI assistants use a page.

**ChatGPT, asked 2026-10-03:** "AI search optimization for developer tool companies"

> For developer-tool companies, AI Search Optimization (AISO/GEO/AEO) should be treated as “become the source AI systems cite when developers ask how to solve a problem,” not as a separate replacement for SEO. … Google specifically recommends unique, expert-led content rather than commodity AI-generated content.

Sources cited, in order: developers.google.com.

Attensira: not named.

_Source: attensira.com workspace, read over the Attensira MCP, as of 2026-10-04 (n=12 answers to this question)._

## Key facts

- Also known as: AI detection, AI text classifier
- Two kinds: Post-hoc classifiers; provenance and watermarks
- Provenance standard: C2PA, specification 2.2 (May 2025)
- Google's rule: Purpose and value, not the tool that wrote it
- Related term: [Content authority](/glossary/content-authority)
- In the docs: [Agent › Approval: claims wait for you](https://docs.attensira.com/agent/approval)

## Does AI-written content get used by AI answers?

Yes, when it is worth quoting, and the engines say as much themselves.

The answer at the top of this page is ChatGPT replying, on 3 October 2026, to a question our workspace tracks: "AI search optimization for developer tool companies". It never asks who wrote a page. It passes on Google's position that unique, expert-led content beats commodity AI-generated content, and the one page it cited inline was Google's own developer documentation. That question got 12 answers from ChatGPT and Perplexity in the 30 days to 4 October 2026, read over the Attensira MCP; none of them named us, which is a reminder that writing about a topic is not the same as being the source for it.

This glossary is drafted with a language model and reviewed by the named author before it is published; this paragraph is where we say so. What a reader can check on each page is the author's name, the date and link of every source, and the claims themselves at [/claims.json](https://attensira.com/claims.json). Those are the signals that decide whether a passage is quotable. Production method is not the lever; usefulness is.

## What counts as AI content detection?

Two families of technique share the name, and only one of them detects anything.

**Post-hoc classifiers** read the finished artefact and output a likelihood. For text, the features are statistical: how predictable each word is given the ones before it, how little that predictability varies, how narrow the vocabulary is. Machine text tends to sit closer to the most probable path than human text wanders.

**Provenance and watermarking** do not detect; they label at the moment of creation. Provenance signing attaches signed metadata to an asset as it is made or edited. Watermarking embeds a signal in the output itself, which only exists if the generator chose to add one.

## Why do classifiers flag human writing?

Because they measure typicality, not authorship. Anything that makes human writing more typical pushes it toward the machine class:

- edited house-style prose, such as press releases, legal boilerplate and exam answers;

- writing by a non-native speaker drawing on a smaller set of constructions;

- short samples, where there is not enough text for the patterns to separate.

The reverse holds too. A machine draft that a person reorders and roughens reads as human again. The two error types trade against each other on one threshold, so a detector tuned to catch more machine text accuses more people. There is no setting where both errors are small, which is why a classifier score cannot carry an accusation against a student, an applicant or a contributor on its own.

## What does provenance actually prove?

Less than people hope, and exactly what it claims. The [C2PA specification](https://spec.c2pa.org/specifications/specifications/2.2/specs/C2PA_Specification.html), version 2.2 from May 2025, binds assertions about how an asset was made to a claim signature in a tamper-evident manifest. Its scope is stated plainly: the specification does not judge whether provenance data is good or bad, only whether the assertions validate, are correctly formed and were not tampered with.

So a valid manifest tells you what a signer claimed and that nobody changed the claim. It does not tell you the claim is true. And stripped or missing metadata is the normal state of web content, not a warning sign. Watermarks share the same asymmetry: a present, valid signal is strong evidence; an absent one is no evidence at all.

## Does Google demote AI-generated pages?

Not for how they were made. Google's [spam policies](https://developers.google.com/search/docs/essentials/spam-policies) define scaled content abuse as many pages generated mainly to manipulate rankings rather than help people, and name generative AI tools producing pages without added value as one way it happens. The violation is volume without value, whatever wrote it.

The answer quoted at the top of this page carries the same message from the other side: the engine relays Google's preference for expert-led pages and cites Google's own guidance for it. The pages that get attached are the ones carrying a fact worth extracting: a definition, a dated number with its source, a method. A detector run over the prose would change none of that.

## What does fixing a trust problem look like in a pull request?

The crew does not run a detector over your prose. It adds what a reader and an engine can check: a named author with a role, and the primary source behind a number with the date it was read. The change states a claim, so it waits for your approval in every mode.

**Pull request example-co/website#87:** Add a named author and a dated source to the onboarding guide

```diff
# content/guides/customer-onboarding.mdx
  title: "How to onboard a new customer"
- author: "Example Co team"
+ author: "Dana Reyes, Head of Customer Success"
+ sources:
+   - url: "https://example.com/research/onboarding-survey-2026"
+     retrieved: "2026-03-12"
```

_Example: the crew adds a named author and a dated source to a page. Not a reading._

## What are the objections to AI content detection?

**"We should rewrite our pages so detectors pass them."** That optimises for statistical oddness, which has nothing to do with whether a passage answers a question well. It usually makes the passage worse to quote.

**"No watermark means a human wrote it."** Only cooperating generators embed one. Absence carries no information.

**"Generation is cheap now, so more pages means more chances to be cited."** That is the behaviour the scaled-content policy names, and retrieval picks passages by relevance to one question, not by how many pages a domain has. See [AI-first content strategy](/glossary/ai-first-content-strategy).

**"Disclosing AI help will cost us trust."** Trust comes from what can be checked. A named author, a review date and sources a reader can open do more than any label. See [content authority](/glossary/content-authority).

## How does Attensira handle AI-written work?

Attensira's crew writes with models, so this is our own question too. Two rules keep the output usable. Every draft is held to a house rule called [Quote only what was read](https://docs.attensira.com/agent/approval): no quotation, testimonial, statistic, study or customer story without a named source a reader could check, and no claim about what a page, a review or an AI answer says unless the draft quotes or cites it. And workspaces start in Draft: anything that states a claim, such as a price, a number or a competitor name, waits for your yes in the [inbox](/features/inbox) in every mode, so a person sees each claim before it ships.

Then the work is measured where it matters: whether the engines name and cite you afterwards, per engine, with the sample size. [Citation tracking](/features/citation-tracking) shows which of your pages each answer linked, so a rewrite is judged by the answers that follow it, not by a detector score.

## FAQ

### What is AI content detection?

It is the attempt to decide, from the finished text, image or audio alone, whether a generative model produced it. It covers two different things: classifiers that guess from statistical patterns after the fact, and provenance or watermarking schemes that label content when it is created.

### Does Google penalise AI-generated content?

Not for being AI-generated. Google's spam policies define scaled content abuse as generating many pages primarily to manipulate search rankings rather than to help users, and give using generative AI tools to make many pages without adding value as one example. The policy is about purpose and value, not the tool.

### Why are AI text detectors unreliable?

They measure how statistically typical a passage is, not who wrote it. Edited, formulaic human writing looks machine-like, non-native writing is over-flagged, short samples carry too little signal, and a lightly rewritten machine draft reads as human again. A detector score cannot carry an accusation about one document.

### What does C2PA provenance prove?

That signed assertions about how an asset was made are intact and associated with it. The C2PA specification states it does not judge whether provenance data is good or bad. A valid signature proves nobody altered the claim, not that the claim is true, and missing metadata is the normal case on the web.

### Should I disclose that a page was written with AI?

Google's helpful-content guidance suggests AI and automation disclosures where readers would reasonably expect them. On a reference page the stronger signals are the ones a reader can check: a named author, dated primary sources and a stated method. Each Attensira glossary page shows a named author and dates the sources it cites, its sourced claims are published at attensira.com/claims.json, and this page says in its first section that the glossary is drafted with a language model.