‘Very disturbing things’: Google’s AI phone editing tools draw concern
Google has unveiled a new AI photo editing feature for its Pixel 9 series of phones called Reimagine that has turned heads among tech experts.
The feature allows users to select any portion of a scene captured on the camera and type in a text prompt. Within seconds, the AI will generate a photorealistic image in the selected space, with convincing lighting and shadow effects.
Google is not the first company to announce new photo editing tools amid an AI development gold rush. Apple revealed an “Image Playground” feature on new iPhones featuring Apple Intelligence debuting this autumn.
However, in an article on The Verge, the team was able to generate a number of concerning fake images with “very little effort” using Reimagine.
‘Huge creative potential’
“We got it to generate some very disturbing things,” Verge reviewer Allison Johnson wrote.
“Some of this required some creative prompting to work around the obvious guardrails; if you choose your words carefully, you can get it to create a reasonably convincing body under a blood-stained sheet.”
The Media Leader has not independently tested the Reimagine feature, which is the latest development as part of Google’s “Magic Editor” toolkit.
Google rolls out AI Overviews to six more markets including UK
While the images created are not dissimilar to what any Photoshop aficionado could create in a reasonable amount of time, the feature being freely accessible to Pixel 9 users and fast to generate such realistic-looking image edits would likely make the production of fake images even more common.
Sarah Salter, head of innovation at WPP agency Wavemaker UK, told The Media Leader she was “impressed” by Reimagine and “as an innovator, see the huge creative potential.”
But she added the feature “has the potential for misuse on a mass scale as it far outpaces where the world is with legal and ethics.”
Google: focused on a ‘helpful editing experience’
Software specialists have also warned the tool would open to abuse by bad actors seeking to spread disinformation.
Sebastiaan de With, an app developer who designed a competitor camera app, Halide, called the development “deeply disturbing” in a post on Threads. “Does anyone working at Google ask themselves if they’re working for the benefit of humanity or is it just made ‘because we can’?” he asked.
When asked whether Google had sufficient guardrails in place to counteract potentially dangerous uses for the photo editor, a spokesperson for Google told The Media Leader that photos that have been edited with Magic Editor “include metadata built upon technical standard from [news media standards body] IPTC,” implying provenance is possible to track.
However, that metadata does not include a watermark. Meaning, as Johnson pointed out, that provenance can currently be circumvented by simply taking a screenshot of the AI-edited image and sharing that online instead.
The Google spokesperson continued: “Our work on Magic Editor is guided by our AI Principles, and we’re focused on providing a helpful editing experience that is informed by user feedback so we can learn, improve, and innovate responsibly on AI together.
“We have clear policies and Terms of Service on what kinds of content we allow and don’t allow, and build guardrails to prevent abuse. At times, some prompts can challenge these tools’ guardrails and we remain committed to continually enhancing and refining the safeguards we have in place.”
Will creative benefits outweigh costs?
While a commitment to continually refining guardrails is necessary, Salter added she was concerned the feature was not “rolled out to a select group of people as a test”.
Instead, the tool is being directly integrated into the new series of Pixel devices that were released to market this week. Meaning, as Salter put it, “millions of users who may not fully understand the ethical implications are able to manipulate images.”
She continued: “Launching this tool at a time when trust in digital media is already low will undoubtedly further undermine trust and challenge people’s ability to distinguish between truth and reality.
“Let’s be clear, however, this isn’t a new issue. This is a macro trend of AI tools and technologies launching faster than our ability to regulate or understand their broader impact.”
On the other hand, Hamid Habib, chief experience officer at Havas Media Group, suggested that the creative benefits of democratised AI tools could outweigh the potential costs.
“Am I worried about AI photo-editing software on the Pixel? No, not really,” he told The Media Leader.
“It’s a great phone and that’s great tech. People should be allowed access to it. The more democratised and easier to use this is, the more creative and innovative we become. Also, I suspect peak everyday folk using generated ‘fake AI’ like the Pope in an East 17 puffer jacket or Donald Trump running away from police trying to arrest him has had its moment.
“People write all kinds of stupid shit in social media today. And of course, people will post some stupid fake pictures in the future. It’s no different.
“Should we be concerned that, at a more serious level, deep-fakes might be used to manipulate the news agenda. Yes, absolutely. But I doubt it’s someone playing on their Pixel [that] is going to bring society to its knees any time soon.”
CMA urged to move against Google following US antitrust decision