Straight answers on AI detection, deepfakes and provenance.
23 threads, 87% answered. Search before you ask. The answer is usually here.
All discussions
- •
Does Google's SynthID or Content Credentials show up when I download an image from Gemini or ChatGPT?
Copywriter. I make header images for client blog posts in Gemini and sometimes ChatGPT. A client asked me last week whether those images "have the AI watermark...
1reply - ✓
How do I verify a viral video is real before my newsroom runs it?
Photo editor at a mid-sized daily here. Our verification workflow for stills is mature: reverse image search, metadata when we have the original, geolocation...
8replies - •
Can a teacher prove I used AI if I have my Google Docs version history?
My daughter is 16. Her history teacher told her an essay "came back 80% AI" and that she has to meet with the head of department next week. She swears she...
10replies - •
Do AI detectors work on code (GitHub Copilot, programming assignments)?
I TA a writing course, but this term I also picked up grading for an introductory Python course and I have run into something I do not know how to think about...
6replies - ✓
Why does Grammarly (or spell-check and editing) make my essay score as AI?
I am a second-year international student and English is my third language. For every essay I write the first draft myself, then I go through it with Grammarly...
7replies - ✓
What AI detector do colleges and teachers actually use, if any?
Most institutions that check for AI do it through Turnitin, because it already sits inside their learning platform. Canvas and Google Classroom have no AI...
1reply - ✓
Do you have to disclose AI-generated content, and who requires labels?
Increasingly yes, depending on where you publish and what the content is. The EU AI Act makes disclosure of deepfakes a legal obligation, YouTube and TikTok...
1reply - ✓
Is it illegal to make or share a deepfake of someone?
Sometimes, and the law is tightening fast. Sexual deepfakes of real people are now criminal in the United States, in England and Wales and in several US...
1reply - ✓
Does Turnitin check for AI, and what does the instructor actually see?
Yes, if the institution licenses and enables it. The instructor sees a percentage of the prose sentences the model thinks were AI-generated or AI-paraphrased...
1reply - ✓
Why do GPTZero, ZeroGPT, QuillBot and Grammarly give different AI scores?
They disagree because they are answering different questions with different models, different training data and different cut-offs. A "70%" from one tool is a...
1reply - ✓
Is using a paraphrasing or humanizing tool academic misconduct?
It depends on what the assessment allows and whether you disclosed it. Most universities now say any AI use in assessed work is unauthorised unless the...
1reply - ✓
How accurate are AI voice detectors in real conditions?
In controlled benchmarks the best systems miss only a few percent of fakes, but accuracy collapses on audio from unseen generators, phone codecs and real-world...
1reply - ✓
How do you tell if a voice recording or call is AI cloned?
You mostly cannot tell by ear, and neither can the detection tools with certainty. The dependable defence is procedural: hang up, call the person back on a...
1reply - ✓
Are AI image detectors reliable, and what do the tests show?
The best commercial detectors score very well on images from generators they were trained on, and much worse on new generators, edited images and adversarial...
1reply - ✓
Do AI humanizers and paraphrasers actually beat AI detectors?
Often, yes, at least for a while: published attacks show that paraphrasing AI text sharply cuts detection rates on most style-based detectors. The caveat is...
1reply - ✓
Is GPTZero accurate compared with Turnitin, Copyleaks and Originality.ai?
GPTZero is accurate enough to flag obvious, unedited chatbot output and not accurate enough to prove anything on its own. In independent tests it has ranged...
1reply - ✓
How can you tell if an image is AI generated?
Usually you cannot tell with certainty from the pixels alone. The reliable route is provenance: Content Credentials or a SynthID watermark can positively...
1reply - ✓
Does watermarking AI-generated text actually work?
It works when the model provider builds it in and the text is not heavily edited, which makes it far more reliable than guessing from style. It does nothing...
1reply - ✓
What is C2PA, and do Content Credentials actually help?
C2PA is an open standard for attaching signed, tamper-evident information about where a piece of media came from and how it was edited. Content Credentials are...
1reply - ✓
Can people reliably spot an AI-generated face just by looking?
No. In controlled experiments, people shown modern AI-generated faces did no better than chance at telling them apart from photographs of real people, and they...
1reply