Mira Murati, CTO of OpenAI, mentioned the work on this technology during Tuesday’s Tech Live event organized by OpenAI Wall Street Journal.

The tool – which is still being tested internally – would determine whether an image of DALL-E 3 has been created (new window) with an accuracy of almost 99%said the technical director on stage.

However, Murati did not provide any information about a release plan for the detection tool but promises to provide more information soon.

There are other tools

Google has already announced the implementation of new features (new window) This makes it easier for Internet users to determine whether a photo was taken by you artificial intelligenceAI. This includes adding a button About the picturethat provides contextual information about its creation, e.g. B. when it was uploaded or whether it was ever identified as fake by news outlets.

Images generated by the softwareartificial intelligenceAI Midjournay or even with the image database Shutterstock are also marked by the Californian giant.

OpenAI recently removed ChatGPT (new window) a small mention: “ Regenerate reaction » (regenerate an answer) at the end of the generated texts, which made it easier to recognize essays created by theartificial intelligenceAI and put it online as is.

The tool also reworded certain sentences, e.g as’artificial intelligenceAIor as a language model ofartificial intelligenceAI to adopt more banal formulas, according to them Decryptor.

The company announced this in January 2023 (new window)by launching a detection tool that it was It is impossible to reliably recognize all written textartificial intelligenceAI.