With the constant rise of nefarious endeavours of generative AI in fake impersonations and crime, Google has taken measures in its Google Magic Editor to prevent users from potentially breaking laws with it on their Pixel phones. According to a study done by Android Authority, they have found that the tech giant has adopted enough guardrails to prevent users from abusing Google Magic Editor.
For example, Android Authority highlighted that Google Photos v6.60 includes code around various error messages that users would see when performing a few prohibited Google Magic Editor edits, such as in anything personally identifiable. How is Google Magic Editor preventing it, and what are the things prevented? Continue reading below.
Related: AI Copyright: Google’s Ambitious Promise to Protect Users in Lawsuits
Google Magic Editor Will Not Allow Edits of Photos of ID, Human Faces, Body Parts and More
As we all know, with Google Magic Editor, we can harness all its generative AI potential to perform complex image edits with ease. However, Google has placed restrictions and boundaries on where people should draw the line. For example, Android Authority discovered that the tech giant had placed strings of error codes pertaining to Magic Editor that would pop up when trying to edit any of the following:
- Photos of ID cards, receipts, and other documents that violate Google’s GenAI terms
- Images with personally identifiable information
- Human faces and body parts
- Selections that are too large for the app to process
Thus, when you try to use Google Magic Editor to edit such pictures with a compatible Pixel device, Google Photos pops up a window to let you know it can’t complete the action since it may violate its terms of use.
Another experiment done by Gizmodo confirmed this case, as the writer mentioned: “I tried this out on my Pixel 8 with the many photos I have of my driver’s licence and Passport. I selected only the part of my headshot for Magic Editor to play with, and it outright refused via a dialogue box, saying that Magic Editor couldn’t complete action on a photo of my card.”
However, it is not foolproof, as we will highlight below.
Google Magic Editor Still Needs Improving
While Google Magic Editor is relatively new since Google first announced Magic Editor at Google I/O 2023 in May, there are still flaws worth improving. For instance, despite the latest update to prevent edits of anything personally identifiable, Android Authority has found loopholes through the security.
According to the same study, they discovered that while Google Magic Editor may block plenty of these edits, some may go through. For example, when they tried editing faces and IDs on an older Google Photos version, Google prevented the edits with a generic error message. However, there was no security pop-up when they tried editing invoices.
Moreover, they have also discovered a code within Google Magic Editor that suggests the arrival of text-prompt-based generation abilities. As such, they should take proper measures in improving the guardrails in refusing to generate images based on some words, as you can technically remix a photo now using Magic Editor on your Pixel 8/Pixel 8 Pro and make an entirely new image without any additional context.
Also Read: The New Nightshade Data Poisoning Tool Lets Artists Fight Back Against Generative AI
Google Is Taking the Correct Measures, but It May Not Affect the Trepidation of AI Editing
Despite Google’s hard work in ensuring a more secure Photos app, it may prove ineffective in easing people’s trepidation of AI-powered editing. With everything considered, Google Media Editor is just a B-tech Photoshop software compared to Adobe’s version.
While both companies are trying their best to provide security measures, it does stop the fear and fact that people are still capable of faking official documentation for nefarious purposes, like swapping out a photo on an ID card to gain access or identity theft.
In fact, the Consumer Sentinel Network Data Book 2022 shows that they took in over 5.1 million reports in 2022, of which 46% were for fraud and 21% for identity theft. Credit card fraud accounted for 43.7% of identity thefts, followed by miscellaneous identity theft at 28.1%, which includes online shopping and payment account fraud, email and social media fraud, and other identity theft.
Therefore, the measures taken by Google may have made it easier to spot forgery if you looked at it for a few seconds, it may change little to quell the fears of AI taking over the world. To know more about what tech giants are doing to combat the threat of AI, follow our Facebook and Instagram pages for daily updates.