Question: Do you know of a model-agnostic detector that can identify AI-generated text, even if it's been edited or paraphrased by humans?

If you want a model-agnostic detector that'll flag AI text, including paraphrased or edited text, GPTZero is a top option. It's geared to detect text produced by large language models like ChatGPT, GPT-4 and Gemini. The service includes Deep Scan for more detailed analysis, Source Scanning for plagiarism detection and Origin Chrome Extension for browser integration. It also offers several pricing tiers, including a free BASIC plan, so you can find something that works for your needs and budget.

Another option is Ghostbuster, which uses a group of smaller language models to analyze documents and estimate their origin. It performs a structured search on the characteristics of multiple models to estimate the origin of the text, though its results are only as good as the models it's trained on. It can be used to detect AI text in academic and professional settings.

If you want a simple interface and high detection accuracy, check out AI Detector. The tool gives you immediate detection results and a breakdown of the percentage of AI used in each sentence. It's geared for students, teachers, bloggers, researchers and businesses that want to ensure authenticity and high quality.

Also, AI Detector gives you a probability score to estimate how likely it is that AI was used in your text. It's particularly good at detecting output from the most widely used AI tools, with an 85% detection rate for text from GPT 3.5 and GPT 4. It's good for content creators, teachers and digital marketers who want to ensure their content is original and optimized for search.