Kalakshar US

YouTube will start using machine learning to determine whether viewers should be on a teen account. The company said it plans to start using this AI application on a subset of US users in the coming weeks for a trial before rolling it out to the rest of the market. The tool will assess user behaviors including the types of videos being searched for, the categories of videos watched and how long the account has existed. When an account is deemed by machine learning to belong to a teen, YouTube will disable personalized advertising, activate digital wellbeing tools and add other safeguards for its younger users.

Since introducing supervised teen accounts in 2021, YouTube has continued to add protections aimed at its younger users. The company noted that it has been using this machine learning approach “in other markets for some time, where it is working well.”

YouTube isn’t the only service to leverage an AI tool to catch users who might be lying about their age. However, it’s putting the burden of correcting false positives on its users: “If the system incorrectly estimates a user to be under 18, they will have the option to verify that they are 18 or over, such as using a credit card or a government ID.” Meta has a similar tool it began rolling out across Facebook, Instagram and Messenger earlier this year, but it lets people simply change their settings if the AI tool incorrectly guesses their age.

This article originally appeared on Engadget at https://www.engadget.com/entertainment/youtube/youtube-is-turning-over-age-verification-to-ai-185634150.html?src=rss 

OpenAI is rolling out a new Study Mode the company says is designed to give students a better understanding of complex topics. Like Claude’s Learning Mode, which Anthropic introduced in April, Study Mode will see ChatGPT adopt a Socratic approach to conversations. Rather than answer a question outright, the chatbot will attempt to guide the user to their own solution, starting with questions that allow the system to calibrate its responses to their objective and understanding. Conversations then unfold using a “scaffold” structure, which means ChatGPT will slowly roll out information so as not to overwhelm the user with more information than they’re ready to digest.

OpenAI says it developed Study Mode in collaboration with teachers, scientists and pedagogy experts. Rather than running on an entirely new model, the tool is powered by a series of custom system instructions.

“We chose this approach because it lets us quickly learn from real student feedback and improve the experience — even if it results in some inconsistent behavior and mistakes across conversations,” said OpenAI. “We plan on training this behavior directly into our main models once we’ve learned what works best through iteration and student feedback.”

Notably, OpenAI isn’t making Study Mode available only to ChatGPT Edu users. Instead, the company is first rolling out the feature to logged in Free, Plus, Pro and Team users. Edu subscribers will gain access in the “next few weeks.” 

It will be interesting to find out how many students end up actually using Study Mode, as a toggle allows you to easily turn the feature on and off. And as a recent New York Magazine article vividly detailed, AI cheating is a major problem at US colleges. For its part, OpenAI says it plans to work on making Study Mode more engaging and useful to students. The company is exploring how to offer deeper personalization through the tool, as well as ways to offer goal setting and progress tracking across conversations.

This article originally appeared on Engadget at https://www.engadget.com/ai/chatgpts-study-mode-will-guide-students-to-an-answer-stey-by-step-180614172.html?src=rss 

NotebookLM, the Google research tool that gained notoriety for its AI-generated podcasts, is introducing a feature called Video Overviews. As the name suggests, this tool automatically creates videos on requested topics and pulls data from a user’s uploaded images, diagrams, quotes and documents.

It’s rolling out right now, but there are some caveats. These overviews are only available in English, though Google says more languages are on the way. Also, the tool doesn’t make videos in a conventional sense. It creates slideshows with narration.

Despite this limitation, the company says the feature is “uniquely effective for explaining data, demonstrating processes and making abstract concepts more tangible.” Google also says it’ll be expanding the toolset in the future, so it won’t always be just a slideshow machine.

Video Overviews do currently offer some handy playback options. There’s the ability to skip back and forth by 10 seconds and adjust the playback speed. The company is also updating NotebookLM’s Studio tab. It’s getting a visual refresh, which rolls out “over the next few weeks” to all users.

Google has certainly been busy iterating on the platform these past few months. It released an official NotebookLM app back in May and began offering curated “featured notebooks” earlier this month. This lets users experiment with the platform with pre-approved topics like William Shakespeare, so newbies won’t have to actually upload anything. Audio-only overviews are also now available in over 50 languages. 

This article originally appeared on Engadget at https://www.engadget.com/ai/google-adds-video-overviews-to-notebooklm-181524866.html?src=rss 

After kicking off its PS Plus 15th anniversary celebrations last month, Sony has announced the PS Plus Monthly games for August. From August 5, members of all of the service’s tiers will be able to add Lies of P (PS5, PS4), DayZ (PS4) and My Hero One’s Justice 2 (PS4) to their libraries.

The headliner is undoubtedly Lies of P, the surprisingly great Pinocchio-flavored soulslike from 2023. South Korea-based studio Neowiz’s twisted take on the oft-reimagined children’s novel stood out in what is now an overwhelmingly crowded genre, thanks to its stylish Belle Epoque-era setting and slick combat. There aren’t many better FromSoft tribute acts around. The post-apocalyptic survival game DayZ launched on PS4 with not-exactly-glowing reviews in 2019, but is still getting fairly regular updates in 2025. Meanwhile, 3D arena fighter My Hero One’s Justice 2 is worth checking out for fans of the My Hero Academia manga on which it’s based.

As it continues to mark 15 years of its Game Pass-rivalling subscription service, Sony has announced that from August 5, members can download a set of anniversary avatars. Themed around the likes of Cyberpunk 2077, Hogwarts Legacy, Diablo IV and God of War Ragnarok, as well as cult retro favorite Twisted Metal, these are also available to subscribers on all tiers.

As for last month’s games, you have until August 4 to add Diablo IV, The King of Fighters XV and Jusant to your game library.

This article originally appeared on Engadget at https://www.engadget.com/gaming/playstation/lies-of-p-leads-augusts-lineup-of-ps-plus-games-165600067.html?src=rss 

Sex toy company Lovense is leaking the email addresses of its app users and allowing account takeovers without asking for a password, according to a security researcher. As reported by TechCrunch, BobDaHacker, who describes themself as an ethical hacker committed to exposing and reporting security vulnerabilities, published an extensive report in which they accuse Lovense of failing to fix a serious bug it was first made aware of in 2023.

According to the hacker (and later verified by TechCrunch), Lovense allows any username to be turned into their email address with the right know-how, a flaw they initially discovered after muting someone on the app. With their access to Lovense’s API, they were able to obtain the emails associated with any public username in less than a second when running the modified request process through an automated script. They noted that the vulnerable nature of these accounts is “especially bad for cam models” who use the Lovense platform for work, and may share their usernames for these purposes.

The researcher also realized that with a user’s email address (either one you already know or one obtained using the aforementioned disclosure bug), they could generate auth tokens that allowed them to take over the associated account without a password. This allegedly worked for the Lovense Chrome Extension and Lovense Connect app, as well as the company’s Cam101 and StreamMaster software — and even admin accounts.

BobDaHacker said they initially reported the bugs to Lovense with assistance from the sex tech hacking project The Internet Of Dongs in March 2025, and received $3,000 in total for flagging them via the HackerOne security platform. After a series of interactions with Lovense representatives, they were told in early June that the account takeover bug had been fixed during the previous month, which the researcher claims is not true. Regarding the email disclosure flaw, Lovense said in a statement printed by BobDaHacker that it could take up to 14 months to fix the issue, as a faster one-month fix would “require forcing all users to upgrade immediately,” which it said would “disrupt support for legacy versions.”

The researcher went on to say that they were contacted by a Twitter user who claimed to have found the same account takeover bug as far back as 2023, and were told shortly after reporting it to Lovense that the bug had been resolved, which wasn’t the case. They said a patch eventually fixed their method, which used an HTTP endpoint to convert a username into an email address, but that it wasn’t rolled out until early 2025. BobDaHacker said they had requested comment from Lovense but at the time of writing had not received one.

This isn’t the first time Lovense users have stumbled upon privacy concern bugs. In 2017, a Redditor discovered that the Lovense app, which allows users to control their sex toys remotely, was recording audio without their consent and saving it to their phone. A commenter on the Reddit post, who claimed to be a Lovense representative, called the recordings a “minor software bug” that affected the Android version of the app and said at the time that it had been fixed in an update.

This article originally appeared on Engadget at https://www.engadget.com/cybersecurity/a-lovense-security-flaw-may-be-letting-people-take-over-accounts-without-a-password-160528730.html?src=rss 

Google is updating AI Mode on desktop this week with the ability to process images, so you can ask it detailed questions about the pictures like you already can on mobile. In the coming weeks, the company is also adding support for PDF uploads on desktop, which could help you digest lengthy course or work materials. You can ask AI Mode to summarize the documents for you and ask follow-up questions that it will then answer by cross-referencing the materials you uploaded with information available on the web. Google says AI Mode’s responses will also include links to its references that you can visit in order to dig deeper. AI Mode will support additional file types for upload, including ones straight from your Google Drive, in the coming months as well. 

In addition to PDF upload support, Google is also rolling out a new Canvas feature that you can access if you’re enrolled in the AI Mode Labs experiment in the US. You can use Canvas to consolidate all relevant information about a specific topic or for a specific purpose in a side panel that updates as you ask AI Mode more follow-up questions. If you’re traveling, for instance, you can ask AI Mode to make you an itinerary and click the Create Canvas button. You’ll be able to keep refining the itinerary with more questions, and you can always leave it alone for a while and come back to it later. 

AI Mode’s Search Live is also getting video input on mobile this week, a feature Google announced at I/O 2025, after voice input arrived in June. To be able to access video input, you’ll have to open Lens in the Google app and tap the Live icon before asking questions on what the camera sees. When Google revealed the feature during its annual developers’ event, it said you could point the camera at a math problem, for example, and ask Search to help you solve it or to explain a concept you’re having trouble understanding. 

Finally, with Lens in Chrome, you’ll be able to ask AI Mode what’s on your desktop screen. The company will roll out an “Ask Google about this page” dropdown option in the address bar “soon.” When you click on it, AI Mode will create an overview with key information on what’s being shown on your screen, whether it’s a web page or a PDF. 

Update, July 29 2025, 12:29PM ET: This story has been updated, as Google shared after the article was published that the image and PDF upload features will be available wherever AI Mode is available, not just in the US. That means those uploads will come to the US, India and the UK.

This article originally appeared on Engadget at https://www.engadget.com/ai/google-is-bringing-image-and-pdf-uploads-to-ai-mode-160050646.html?src=rss 

Netgear has launched its most affordable Wi‑Fi 7 mesh system yet, the Orbi 370. The headline price is $350 for a router and two satellites, which is extremely competitive. A quick look at the specs, however, reveals several trade‑offs that make the lower cost possible.

The company says the Orbi 370 is “designed for customers with moderately demanding Wi‑Fi needs,” claiming the system’s total bandwidth is 5 Gbps. However, since it carries only a 2.5‑gigabit internet port, the maximum speed to a single device is limited to 2.5 Gbps.

The Orbi 370 is strictly dual‑band, with only 2.4 GHz and 5 GHz radios — there’s no 320 MHz or 6 GHz band options. This means the Orbi 370 is missing one of Wi-Fi 7’s biggest advantages in most setups.

Households with a two‑gig connection or less likely won’t miss the extra headroom, and a mesh Wi‑Fi system remains one of the best ways to solve dead spots around your home. Netgear’s flagship model, the Orbi 970, offers 27 Gbps total bandwidth, with a 10‑gig internet port and a quad‑band setup. That model comes with a price tag of roughly $2,000 for a router and two satellites and would likely be overkill for most households.

This article originally appeared on Engadget at https://www.engadget.com/home/smart-home/netgear-launches-its-most-affordable-wi-fi-7-mesh-system-163234180.html?src=rss 

Elgato just introduced a novel new webcam with a pro-grade feature. The Facecam 4K allows users to mount lens filters for unique effects. This is the first time a webcam has offered this functionality.

The camera integrates with any 49mm lens filter, which opens up a world of potential. Folks could attach, say, a circular polarizing (CPL) filter to reduce reflections from glasses or go crazy with cinematic effects like diffusion. The unit ships with a free CPL filter to get started, as long as you buy the camera from the official Elgato store.

As for traditional specs, it records in 4K at up to 60 FPS and uses Sony’s Starvis 2 CMOS sensor with a f/4.0 aperture and fixed focus. It connects over USB-C and weighs just 0.26 pounds.

The Facecam 4K pairs with Elgato’s Camera Hub software, which offers manual control over a variety of settings. These include ISO, exposure, shutter speed, pan, tilt and zoom. It also works with the company’s Prompter hardware, though that requires a replacement backplate. The company is throwing in the dedicated backplate with new Prompter purchases. For the uninitiated, the Elgato Prompter is a nifty teleprompter designed for streamers.

The Facecam 4K even pairs with the Switch 2, which could make for some truly cinematic Mario Kart World reaction shots. Just update the firmware to version 2.32 or later and connect directly to the console.

Elgato’s latest webcam is available right now and costs $200. This unit is similar in features to the pre-existing Facecam Pro webcam, but that one costs $300 and doesn’t let people swap out lenses. It does, however, include a better sensor.

This article originally appeared on Engadget at https://www.engadget.com/entertainment/streaming/elgatos-new-webcam-can-mount-lens-filters-163656839.html?src=rss