CiteOwl
CiteOwl
Blog/Using AI honestly

Using AI honestly

How AI detectors work, what a Turnitin AI score means, where universities draw the line on AI use, and what to do if you are accused.

Most universities have stopped asking whether students use AI and started asking how. Policies differ by school and often by course, but the line usually falls in the same place. Help with thinking, structuring, and editing is fine. Submitting work you cannot explain is not.

Detection is the part everyone worries about, and the part that moves fastest. Below is where it actually stands, dated, because an answer to this question that carries no date is worth nothing.

What is actually detectable, as of 11 August 2026

The useful question is not "can AI writing be detected". It is "can the specific person marking my work run a specific check today, and what would a hit prove". Those have different answers, and several of them changed in the last two weeks.

The checkCan your marker run it todayWhat a hit would actually prove
A statistical AI detector
GPTZero and the rest
Yes. Anyone can paste text into a free one Very little alone. These measure how predictable text is, so plain, careful writing scores high. A 2023 Stanford study found detectors misclassified more than half of essays by non-native English speakers
Turnitin's AI indicator Only if the institution licenses it and left it switched on, which is not universal. Vanderbilt publicly disabled it in 2023 Turnitin's own published position is a document-level false positive rate under 1%, caveated to documents where it reports over 20% AI writing, with an asterisk marking scores below that as less reliable. The company says the report should not by itself be used to penalise a student
Claude's text watermark No, not yet. Anthropic started marking with EU launches on or after 2 August 2026 and says detection is covered in "forthcoming technical documentation" When it arrives, that text "may have been processed by Claude". Which includes Claude having only edited writing that was entirely yours
Gemini's SynthID text watermark No. The SynthID Detector portal is in limited testing behind an early-tester waitlist The same. A watermark says a model touched the text, not who wrote the argument
C2PA credentials on an image Partly. The public verify page reads them when they survived Only that a signed record made it through. A screenshot destroys it, so absence proves nothing at all
Invisible characters and file metadata Yes, trivially, and this is the one people forget That a passage was pasted from somewhere. Not from where
Your document's version history Yes, if you submitted a shared file or are asked for it A drafting trail. This is the check most likely to help you, which is the whole argument for keeping one
Whether your sources exist Yes. Free, no licence, about a minute per source, by anyone including your marker Everything. A reference that resolves to nothing is not a probability score, it is a fact, and it is what actually ends up in front of a panel

Read down the middle column. Every check that runs reliably today is either about your sources or about your drafting trail. Every check that is about your prose is either statistical and contested, or not yet available to the person grading you. That is not a loophole, it is a planning fact: the effort you spend worrying about the third column is misallocated, and the effort you spend on the last row is not.

What changed on 2 August 2026

Two things landed on the same day and they are connected. Article 50 of the EU AI Act, the transparency article, applies from 2 August 2026. It obliges providers to "add machine-readable marks to enable the detection of AI-generated or manipulated content", and obliges deployers to disclose deepfakes and AI-generated "text publications on matters of public interest without human review or editorial control".

Anthropic's response is visible in its own documentation: Claude models launched in the EU on or after that date "will support machine-readable marking at launch", using an imperceptible text watermark that "may persist through some editing" plus signed C2PA metadata on supported files.

Three things that law does not do, and it is worth being precise because a lot of coverage was not. It does not oblige you to disclose AI use in an essay; that is your university's rule, not Brussels'. It does not hand your marker a detector, and as of today none of the marking schemes above has a public one. And a machine-readable mark is not an authorship finding: Anthropic says plainly that a detected mark means the content "may have been processed by Claude", which is also true of a paragraph you wrote and asked it to tighten.

Where each guide fits

One thing this site will not help with, and says so on the front page: getting AI writing past a detector. Not because of the detectors, which are unreliable in both directions, but because the workflow that survives every check in the table above is the same workflow that produces work you can defend in a five-minute conversation. There is no second method.

20 guides

All topics
Is using AI to write essays cheating?

The honest answer is "it depends on what you use it for." Here is how universities tend to split allowed help from misconduct, and how to stay on the right side.

8 min read
Will AI detectors flag my writing?

What AI detectors actually measure, why they flag plenty of genuine human writing, and what you can and can't read into the score before you hit submit.

8 min read
Can Turnitin detect ChatGPT? What actually happens in 2026

Turnitin can detect ChatGPT sometimes, through a separate AI indicator that misses edited text and wrongly flags honest writers. What it really takes to be in the clear.

10 min read
Does AI text have a hidden watermark?

Anthropic began marking Claude's text on 2 August 2026 and Google has marked Gemini's for a year. What the mark is, how much of it survives an edit, and why nobody grading you can read it.

12 min read
How to check a document for hidden AI watermarks

Three different things hide in a document: invisible characters, file metadata, and a statistical watermark. How to inspect the two you can, and what the third is worth.

9 min read
Do AI images carry hidden watermarks?

A pattern in the pixels survives almost anything. The signed record beside it dies to a screenshot. What that means for a figure in your paper, and what journals actually ask for.

10 min read
Using ChatGPT for an essay: what actually gives it away

What gives a ChatGPT essay away isn't the writing style but fabricated citations, sourceless content, and a missing drafting trail. The honest fix is doing the work, so there's nothing to catch.

12 min read
Do AI humanizers actually work? An honest look

AI humanizers can slip text past one detector for now, but they can't make it reliably undetectable, and the fake citations, lost voice, and integrity rules stay exactly where they were.

12 min read
Your Turnitin AI score: 4% of flagged sentences are human

What the Turnitin AI writing percentage counts, what it cannot see, how it behaves on human writing, and what a score does and does not prove.

9 min read
Accused of using AI on an essay: what to do

What evidence actually helps, what an academic integrity meeting is like, and why the right move differs depending on what really happened.

8 min read
Can Turnitin detect Claude? What the published list says

Turnitin names seven Claude versions it covers, and stops. What that list tells you, and why Anthropic's own watermark is a second system nobody can read.

11 min read
What is a good Turnitin similarity score?

There is no pass mark, and Turnitin says so. What the percentage counts, what your instructor changed before you saw it, and what twelve real universities publish.

12 min read
Does Turnitin detect QuillBot?

Turnitin names QuillBot in its own documentation. What it is really finding, why the one honest study is out of date, and where every circulating percentage came from.

13 min read
17 universities that turned off Turnitin's AI detection

Seventeen institutions with their own announcements, the nine we could not verify, and what the widely repeated 50-university figure actually counts.

16 min read
AI disclosure statements: what to actually write

Your syllabus asks you to disclose AI use and never says how. The six things institutions want, five templates, and which policies would still refuse each one.

14 min read
Can I reuse my own paper? Nine universities say ask first

Nine university policies, quoted. What needs permission, what does not, and why Turnitin matches you against yourself.

12 min read

More on the blog

A draft that shows exactly what the agent did

Free to start. No card needed.

Start writing