Are AI detectors accurate?Not reliably; a score is a clue, not a verdict

TimeBinder adds a layer of protection against AI and human fraud and disputes.

The file stays on your device. Your browser makes a digital fingerprint of the file. TimeBinder records the fingerprint and the time, not the file.

Short answer

Not reliably. AI detectors for text and images can flag human work or miss generated work. A score is a clue, not a verdict. Keep dated copies of your own drafts if you may need to show when they existed. A TimeBind records a fingerprint and a time; it does not detect AI.

This short video explains why an AI-detector score is a clue, not proof, and what to check instead.

AI Image Detector: Can You Trust the Score? (39 seconds).

What detectors measure, and why they err

An AI detector typically reads a finished text or image and estimates how likely it is to have been machine-generated. It produces a score, not a record of how the work was made. Makers and independent researchers have shown these tools can be wrong. OpenAI released its own AI text classifier in January 2023 and withdrew it on 20 July 2023, saying it was “no longer available due to its low rate of accuracy.” At launch, OpenAI reported that the classifier correctly flagged 26% of AI-written text and wrongly labelled human-written text as AI-written 9% of the time on its evaluation. Other tools have different methods and different error rates; none is a verdict.

Image detectors have the same kind of limit: they can miss generated pictures and flag real photographs. For that question, see how to tell if an image is AI generated. A longer comparison of detector scores versus a draft trail is in working drafts and timestamps versus AI detectors. If a publisher’s own detector has flagged your work, see what writers should keep when a detector is involved.

What to keep instead of relying on a score

If you may need to show your process, keep dated copies of drafts from before any AI pass, plus later versions. A TimeBind records a SHA-256 fingerprint of each exact file and a time, made in your browser, with the file not uploaded. It shows that file existed no later than the recorded time and still matches. It does not detect AI, score a passage, or prove you wrote the words. For a simple routine, see draft history and how to reply to an AI accusation.

Make a TimeBind: first one free →

First Pooled TimeBind free, no card, then from US$5. An ASAP TimeBind is US$15.

Frequently asked questions

Why do AI detectors give false positives?

They estimate how predictable or “machine-like” a finished text or image looks. Human writing and real photographs can look that way too, especially if the prose is formal or the photo is clean. A score is not a record of how the work was made.

Can I use a detector to prove a photo or essay is real?

No. A low AI score does not prove a file is human-made, and a high score does not prove it is generated. Use it only as a clue, alongside original files and any labels that are still attached.

What should I keep if I am accused of using AI?

Keep dated copies of your own drafts from before any AI pass, earlier emails or uploads, and a dated fingerprint of those files if you made one. A TimeBind shows timing and a match; it does not disprove the accusation by itself. See how to reply to an AI accusation.

Does a TimeBind replace an AI detector?

No. They answer different questions. A detector estimates whether finished work looks generated. A TimeBind records that an exact file existed no later than a recorded time and still matches. It does not detect AI.

Sources

OpenAI, New AI classifier for indicating AI-written text (launch figures) and the 20 July 2023 withdrawal stating a low rate of accuracy.
TimeBinder, Working drafts and timestamps versus AI detectors.

Related guides and tools

Your file’s fingerprint, anchored to a global network secured by real electricity and cryptography.

Editing a file—including with AI—changes its fingerprint, not its existing network record.

How this works

Your file’s SHA-256 fingerprint, a short code worked out from its exact contents, is made in your browser, and the file itself stays on your device. A cryptographic commitment to that fingerprint, which may be combined with others, is then anchored in a block on the Bitcoin blockchain. Bitcoin is secured by proof of work: computers around the world spend real electricity to add each block, and changing an earlier block would mean redoing that work faster than the rest of the network. Until that block is confirmed, your proof is pending. Once the Bitcoin anchoring is confirmed, you can verify the proof against that recorded commitment. The confirmed record can then be checked; Pooled TimeBinds are checked through TimeBinder, while ASAP TimeBinds can also be checked independently. If you check the file later and its fingerprint still matches, that shows the file existed no later than the recorded time. The record doesn’t show who made the file, whether AI was used, or that its contents are true.

Your filestays on your device
FingerprintSHA-256
Network timestampon Bitcoin

Record a fingerprint of the file you rely on

Add a layer of protection against AI and human fraud and disputes: a dated, checkable record of the exact file you hold.

Open the app and sign in with email or Google.
Add your files. They stay on your computer.
Seal them and download your Proof of Time certificates.

Open TimeBinder →

First Pooled TimeBind free, no card, then from US$5. See pricing.