Has Google given up on guardrails?
Camera Fake
Google's latest AI photo editing tool, “Reimagine,” lets you add or remove objects in photos using text prompts, with the results being “mysterious,” reports The Verge.
The feature, introduced by Google alongside the launch of its Pixel 9 smartphone earlier this month, allows anyone to tweak photos within the camera app, easily making it possible to unintentionally discredit a photo.
Examples shared by The Verge's Chris Welch included a photorealistic image of the aftermath of a bike-car collision, a lion roaming behind a locked gate, and “a mysterious liquid spurting from a Metro-North train.”
Corpses and drugs
Throughout its testing, The Verge encountered few guardrails with the new feature, easily adding “car wreckage, a smoking bomb in a public place, a sheet that appears to be covering a bloody corpse, and drug paraphernalia” to images.
In a statement, a Google spokesperson directed The Verge to its terms of use, which spelled out the specific types of images generated by The Verge, and insisted that the company is “committed to continually strengthening and improving the safeguards we have in place.”
In another test, The Verge uploaded an edited image to an Instagram Story to see if Meta would tag it and notify other users that it was fake, but the system couldn't do so. This isn't all that surprising, given that Meta's “Made with AI” labeling efforts have failed, even going so far as to mistakenly label real photos as AI-generated.
So while photo alteration using generative AI has been around for some time, Google's latest photo editing tool enhances the technology by making it incredibly accessible, setting a dangerous precedent for a future filled with fake images.
But given that former President Donald Trump has already weaponized the technology to attack his rival, Kamala Harris, we're already there.
More on generative AI: Donald Trump says he knows nothing about the AI-generated Taylor Swift image he posted
