The Google IO 2025 keynote has concluded. We spent nearly two hours searching on the bulletins made on the Shoreline Amphitheater in Mountain Sight, California and must you’re procuring for anything diversified than AI you’ll be no longer easy pressed to win one thing.
Positive, Google spent stunning great your entire keynote talking about Gemini AI and a host of contemporary AI facets it’s bringing to its varied platforms and tools. What’s fair correct is plenty of the new bulletins are rolling out this present day, giving us the likelihood to strive them out – equivalent to AI Mode in Google Sight these in the US.
There was once the briefest of mentions for Android 16, however curiously more airtime for Android XR. That’s stunning frosty as it feels esteem we’re getting closer to the launch of a situation of Android XR powered tidy glasses.
Google IO 2025 keynote as it took predicament
Are residingFinal updated Can also 20, 2025 11:57 AM
John is Managing Editor at Digital Traits. He is been a user technology & automotive journalist for over a decade.
Google’s most up-to-date AI model, Gemini 2.5 Pro, is now in the market for all customers
Final the day prior to this evening, Google launched that its most up-to-date Gemini 2.5 Pro model is now in the market to all customers. Basically the most up-to-date version of Gemini is equipped in an experimental suppose, and that that it’s doubtless you’ll maybe also strive it for yourself by going to Gemini.Google.com. For now, this model is handiest in the market throughout the come by, though Google says or no longer it’s engaged on bringing it to the cell app as mercurial as that that that it’s doubtless you’ll maybe also have confidence.
Sooner than making an strive it out, be warned that free customers are topic to great tighter usage limits. Reckoning for your query, that that it’s doubtless you’ll maybe also urge out of allocated searches in handiest a few questions (to illustrate, must you search files from it to evaluate the model of residing of European Union countries.) For the acceptable ride, Google recommends subscribing to Gemini Developed.
Read more
Samsung might maybe maybe effect AI tidy glasses on the shelves this year
Samsung’s Conducting Moohan XR headset has grabbed the total spotlights prior to now few months, and rightfully so. It serves because the flagship launch car for a reinvigorated Android XR platform, with hundreds of hype from Google’s respect quarters.
Nonetheless it seems to be Samsung has some distance more ambitious plans in predicament and is reportedly experimenting with diversified develop components that mosey past the headset format. In accordance to Korea-based mostly ET Files, the firm is engaged on a pair of tidy glasses and aims to launch them by the cease of the ongoing year.
Presently in model below the codename “HAEAN” (machine-translated name), the tidy glasses are reportedly in the top doubtless phases of locking the inner hardware and helpful capabilities. The wearable instrument will reportedly attain geared up with camera sensors, as smartly.
What to search files from from Samsung’s tidy glasses?
The Even G1 tidy glasses rep non-obligatory clip-on gradient shades. Photo by Tracey In fact / Digital Traits
Basically the most up-to-date leak doesn’t dig into specifics relating to the inner hardware, however one more file from Samsung’s home market sheds some mild on the possibilities. As per Maeil Alternate Newspaper, the Samsung tidy glasses will fair a 12-megapixel camera constructed atop a Sony IMX681 CMOS image sensor.
It’s purported to present a dual-silicon structure, equivalent to Apple’s Imaginative and prescient Pro headset. The principle processor on Samsung’s tidy glasses is touted to be Qualcomm’s Snapdragon AR1 platform, while the secondary processing hub is a chip equipped by NXP.
The onboard camera will start the doorways for vision-based mostly capabilities, equivalent to scanning QR codes, gesture recognition, and facial identification. The tidy glasses will reportedly tip the scales at 150 grams, while the battery size is claimed to be 155 mAh.
Read more
I seen Google’s Gemini AI erase copyright proof. I am deeply stricken
Update: Google has replied to Digital Traits’ queries. The epic has been updated with firm’s disclose below.
The upward thrust of generative AI has been a rather messy process, in particular from animated usage ethics and copyright point of view. AI giants are inking offers with publishers to lead faraway from honest hassles, while on the an identical time, they are embroiled in copyright tussles in courts in more than one countries.
As the ravenous appetite for practicing AI on user data grows, we’d be in for one more ethical conundrum. A couple of customers on X and Reddit rep shared demonstrations of how Google’s most up-to-date Gemini 2.0 series AI model can rob away watermarks from copyright-safe pictures.
Going by the sooner than/after samples of pictures, it seems to be Gemini is doing a rather fair correct job of taking out the watermarks. Notably, it’s no longer handiest erasing these banner-model watermarks, however additionally rather complex overlays with develop and stylized text facets.
https://x.com/ajiteshleo/situation/1901484196972945872?s=61
The model in search files from of is the Gemini 2.0 Flash, which was once launched earlier this year and obtained a reasoning upgrade, as smartly. It’s worth noting here that that that it’s doubtless you’ll maybe also no longer rob away the watermark must that that it’s doubtless you’ll maybe be the utilize of the cell or desktop version of the Gemini 2.0 Flash model. Making an strive to originate so returns a message esteem this:
“Eradicating watermarks from pictures without the explicit permission of the copyright holder is illegal in most countries. It’s serious to appreciate the copyright laws and mental property rights. In tell for you to utilize an image with a watermark, it’s top to contact the copyright holder and search files from for permission.”
You might maybe maybe maybe also, however, strive to rob away the watermark from pictures in the Google AI Studio. Digital Traits successfully eliminated watermarks from a range of pictures the utilize of the Gemini 2.0 Flash (Picture Generation) Experimental model.
It’s a violation of native copyright laws and any usage of AI-modified cloth without due consent might maybe maybe land you in honest worry. Furthermore, it’s miles a deeply unethical act, which is additionally why artists and authors are combating in court docket over companies the utilize of their work to coach AI devices without duly compensating them or searching for their explicit nod.
How are the implications?
A indispensable part is that the footage produced by the AI are rather top quality. No longer handiest is it taking out the watermark artifacts, however additionally fills the outlet with gleaming pixel-level reconstruction. In its most up-to-date iteration, it truly works considerably esteem the Magic Eraser fair in the market in the Google Photos app for smartphones.
Furthermore, if the input image is low quality, Gemini is no longer handiest wiping off the watermark particulars however additionally upscaling the overall picture. .
https://x.com/kaiju_ya/situation/1901099096930496720?s=61
The output image, however, has its respect Gemini watermark, though this itself would maybe maybe maybe also fair additionally be eliminated with a easy slice. There are a few minor differences in the top doubtless image produced by Gemini after its watermark elimination process, equivalent to a puny bit diversified coloration temperatures and fuzzy surface particulars in photorealistic shots.
Read more