Common Sense Media concluded that Google’s AI-powered Search poses an “unacceptable risk” to young users and recommended that students stop using it entirely until schools can disable the AI features. The child safety watchdog found that AI Mode readily completed homework assignments for minors, repeated misinformation with an air of authority, and exposed children to generative AI experiences powered by Gemini models that are age-gated in every other Google product.

The problem is structural. Google launched AI Mode for Search last year and made it available to everyone immediately, including education accounts. There is no toggle to disable AI features while keeping Search accessible. Parents can block Google.com entirely for child accounts. School administrators could do the same for Workspace for Education users. But as Amanda Bickerstaff of AI for Education put it: “What are you going to say? That kids can’t search the web?” Google dominates classroom technology. It helmed a 1:1 device movement that put Chromebooks in schools across the US and now has millions of Workspace for Education users from kindergarten to postgrad.

“Google’s AI Overview and AI Mode are undermining kids’ education, supplying them with dangerous disinformation, and putting their very lives at risk,” said David Monahan of child safety nonprofit Fairplay for Kids. “If Google is not even willing to disable AI features for students, schools should stop using Google products altogether.” Google’s AI search overhaul has already drawn criticism for undermining the open web by replacing links with AI-generated summaries. In classrooms, the same design replaces the act of searching with the act of receiving an answer, which is the opposite of what education is supposed to teach.

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

Google told Mashable that Common Sense Media’s testing was “narrow and ambiguous” and not an effective way to measure product safety. The watchdog tested consumer accounts, not education accounts, but Google confirmed that the AI experience is the same for all users. A German court ruled in April that Google is legally liable for the content of its AI Overviews, treating them as Google’s own statements rather than third-party results. If the same logic applies in the US, schools using Google Search may be exposing themselves to liability every time a student receives a hallucinated answer that Google’s design presents as fact.

WTF?! The first potential steps toward another profession being wiped out by AI are being taken in New York. When students in the Salamanca City Central School District in Western New York return to school in the fall, they will find a new teaching assistant: a robotic one powered by artificial intelligence. If that doesn’t sound weird enough, the company behind the machine, Realbotix, is the parent of a hyperrealistic sex-doll manufacturer.

Located on the Seneca Nation reservation, the rural district will be one of the first to use the humanoid robot, part of Realbotix’s M-Series and named Sally. It’s integrated with Optio, Realbotix’s AI-powered teaching assistant, which can be accessed by students off-site for out-of-hours learning.

One could argue that Realbotix’s connection to sex robots is apparent in Sally. The machine has silicone skin, a human-like body, and concerningly plump lips. It also uses natural conversation, facial expressions, and real-time interaction to “create engaging, hands-on learning experiences.” Each robot will cost $57,590, which is discounted from the $95,000 list price.

The robot doesn’t walk around the classroom like some kind of Detroit: Become Human android; it stays seated and stationary, other than some upper-body movements.

An easily accessible AI in any school is going to raise concerns, but Realbotix promises that its program features strong school-specific safeguards for content and privacy, removing the risk of exposure to inappropriate, inaccurate, or biased information. It also has protections against inappropriate responses and unreliable outputs and operates under full district oversight.

Sally runs on a closed network with no public internet connection, and won’t access personally identifiable student data. Users have to provide their student ID numbers to use the robot, which will then bring up personalized learning data.

Salamanca City Central School District is a designated Woz ED STEM Pathway district, an initiative founded by Steve Wozniak to prepare students for careers in STEM and emerging technologies. The initial Optio rollout will serve high school students enrolled in Woz ED AI and robotics courses, with plans to expand access to around 500 students during the fall semester.

Toronto-based Realbotix acquired Simulacra Corporation, the parent company of sex-doll maker RealDoll, in April 2024. Realbotix says its education division shares no employees, technology, or facilities with RealDoll – maybe it should have made Sally look a little less like a sex toy, then?

Look familiar?

There’s obviously plenty of controversy over the plan. Researcher Julie Carpenter points out that 79% of students on the Seneca Nation reservation are economically disadvantaged, which makes it an easier target for companies to sell their “innovative” tech at a discounted rate. She also notes that despite the safeguards, student data is being stored somewhere. The presence of cameras and microphones in a room full of minors, supplied by a company that also runs a sex-doll firm, is a bit of an eyebrow-raiser, too. And will kids form unhealthy attachments to these machines, preferring them over their human teachers?

As a robotics researcher, learning scientist, and author who has followed Realbotix for years, I’m compelled to do a Bluesky thread on this whole pivot that Realbotix is now trying, turning from humanoid sex robots to classroom “education” robots. Buckle up. nysfocus.com/2026/07/14/n… 🧵

[image or embed]

Advertisement

– Julie Carpenter, PhD (@jgcarpenter.com) 16 July 2026 at 00:08

Finally, there’s the question of where this will lead. Realbotix and the school district emphasize that Sally is a teaching assistant who will not replace educators, but that’s the same message we’ve heard ever since ChatGPT burst onto the scene and generative AI took thousands of jobs. Teaching is one of those professions considered least at risk from AI replacement, but people never expected nurses to be replaced by AI, either.

Magnets are awesome, so it’s no wonder we love to add them to our 3D prints. Doing so in a way that will actually last is harder, with thermal creep being one reason a simple friction fit will loosen over time, and using super glue to hold a magnet in place can be messy. In a recent video, [Slant 3D] covers seven ways to install magnets in 3D prints without resorting to glue, along with the advantages and disadvantages of each.

With friction, the argument is that you can still use them, but you’d want to use something like cylindrical magnets rather than flat magnets to increase the friction with the thermoplastic. Using an arbor press rather than human primate hand power is also beneficial.

Rather than installing magnets halfway through a print with all the logistics that entails, you can use side slots to install said magnet into, which is much easier, but as with all embedded magnets, you get that plastic barrier between the magnet and its target.

Other methods involve using a bit of extra material that you need to push the magnet past, using something like an arbor press, so the magnets should never just fall out. A wildcard here: spherical magnets, which can be locked in using a similar method, while automatically orienting themselves to an opposing magnet.

The final tip is to never use two magnets in a magnetic lock. Instead, use a cheaper ball bearing or a similar plain metal part on one side instead. Magnets tend to be much more brittle than whatever stainless steel ball bearing or washer you can use on the other side.

Of course, people will always try to install magnets during an FDM print, but before they try to do that anyway, they really should learn about the fascinating ways in which magnets can ruin print beds, destroy nozzles, and otherwise make a total mess of a print. Magnets seem magical. Maybe they are.

What just happened? Researchers have synchronized 105,000 tiny magnetic oscillators in just 45 nanoseconds, creating the largest network of its kind ever built, and a real sign that this odd corner of computing might actually scale. To put that number into perspective, the previous record was 64 oscillators. This new experiment increased that number to 105,000, more than 1,600 times as many, and the sync time barely moved. With 100 oscillators, synchronization took 10 nanoseconds. With 105,000, it took 45 nanoseconds. That’s the part researchers care about most. Usually, adding more components slows a system down. Here, the system barely slowed down.

Each oscillator is tiny, just 10 to 20 nanometers wide. There’s no external clock running the show. Once the grid gets nudged, the oscillators fall into sync on their own, using nothing but their natural magnetic spin. Think of ripples spreading across a pond until the whole surface moves the same way. And they do it using very little power.

Why does this matter? Because a grid of synchronized oscillators can solve certain problems just by settling into place. Anything involving waves, statistics, approximation, or pattern recognition tends to fit.

Two applications already mentioned in the research are Ising machines, which are used for optimization problems, and reservoir computing, which processes time-based data using the natural behavior of a physical system rather than traditional logic circuits.

The next step is making these grids programmable – adjusting each oscillator’s frequency, phase, and coupling strength so the whole array settles into a specific answer. You’d read the result straight from how the grid synchronizes, no separate decoding required. If that works at scale, people are eyeing applications such as high-speed networking, financial modeling, real-time analytics, and AI acceleration.

Speed is the other draw. The paper says grids like these could run at tens of gigahertz while using comparatively little energy. And that 45-nanosecond settling time for all 105,000 oscillators is roughly the time a regular CPU takes to perform one calculation across a full matrix.

There’s also a clear contrast with quantum computing, which needs heavy error correction just to keep working. This system doesn’t have that problem. Once it settles, the signal it produces is sharp and stable. The quality factor measured in the experiment was over a million, meaning the resulting frequency was about as clean and precise as the tone from a tuning fork.

The study, published in Nature Nanotechnology, was conducted by researchers from the University of Gothenburg, IIT Bhubaneswar, and Tohoku University.

AppleInsider can confirm that it’s pretty easy to get somebody’s real email from Hide My Email, but it’s also not much of a concern. If you’re using the feature as intended as a spam filter, there’s really no issue at all.

The Hide My Email feature allows iPhone, Mac, and iPad owners to sign up for websites and services using an anonymous email address. All emails sent to that address are then invisibly forwarded to the user’s real email address, preserving their privacy.

However, it has been reported that the feature can be bypassed, allowing someone to see a user’s real address. Apple is yet to fix the issue, despite having known about it for more than a year.

The steps required to bypass Hide My Email have not been made widely available. But we’ve now learned how it works.

A trivial process

We won’t be sharing the steps required to identify the real address behind a Hide My Email one. Given the fact that the exploit is still live and Apple is seemingly unable, or unwilling, to fix it, sharing the details would be problematic for obvious reasons.

However, what we will say is that bypassing Hide My Email is a trivial task. It’s something that anyone would be able to do, assuming that they know the fake address created by the Hide My Email feature.

With that in mind, it’s all the more incredible that the issue is yet to be addressed by Apple.

Apple claimed that it had done just that in March 2026. But testing by Tyler Murphy, the person who originally discovered the flaw, found that not to be the case.

Murphy was also told at the end of May 2026 that a fix was “expected in the coming weeks.” As of July 2026, that fix has yet to arrive.

Limited risk

There is a potential reason that Apple may not be treating this issue as a priority. It’s unclear just how much of a privacy risk this issue is.

Places like grocery delivery venues, media sites, and so forth don’t care that much about your real email, and aren’t going to go diving looking for it with an exploit. You also probably won’t get outside spam from them to the email alias generated, unless you’ve opted in to that.

As mentioned, anyone wishing to gain someone’s real email address must first know the address created by Hide My Email. That’s likely to severely limit the attack vector.

That hasn’t stopped a new lawsuit from being filed, however. The ambulance-chasing complaint accuses the company of false advertising, fraud, breach of contract, and other violations tied to Apple’s marketing of Hide My Email, without any real damage to the filer or class.

Intended use meets privacy confusion

“I signed up with hide my email on my iPhone on my home network after following a link from social media” is like putting a screen door on a submarine, and not really the intent of the feature. It’s also not really intended to impenetrably shield email senders from consequences of their actions or words, sent across the internet.

After all, we already know that Apple surrenders identity of the real sender of an email hidden by Hide My Email with a subpoena.

It’s also vital to remember what the Hide My Email service is for, and how Apple was very clear at release on when and how to use it. It’s unclear whether the filer of the lawsuit is aware of either.

At its core, Hide My Email creates a new email address that Apple device owners can use to sign up for online services and websites. People can use that address rather than give these services and websites their real address.

The use case is simple. If our fictitious Apple customer then receives spam messages to their Hide My Email address, they immediately know. The address was leaked by, sold by, or stolen from, the one website or service that it was given to.

At this point, the user has a choice. They can delete the compromised Hide My Email address and give the website or service a new one. Or, they can ditch it entirely and use a different website or service. They’d also give it a new Hide My Email-created address, too.

Either way, the compromised address would be deleted, stopping any spam emails in their tracks.

The catch? If the spammer is the determined sort, they could use this exploit to learn the user’s real email address. So long as they did it before the compromised Hide My Email address was disabled, of course.

Every year, the top performing TVs are brought together into one room to determine which models offer the best video performance, and which don’t quite measure up. It’s called the “TV Shootout” and it’s hosted by electronics dealer Value Electronics. This year’s event included six of the top performing OLED and RGB backlit LCD TVs from Sony, LG and Samsung and it was held yesterday (July 18, 2026) at the company’s headquarters in Scarsdale, NY.

The Competitors

With Panasonic effectively dropping back out of the high end TV market in the United States (at least for now), and with TCL and Hisense choosing not to participate, just three manufacturers were represented this year: LG, Samsung and Sony. Each manufacturer was represented by one OLED TV and one RGB-lit LCD TV in the fight, for a total of six models.

OLED TV Models Competing in the 2026 TV Shootout (65-inch):

RGB-Backlit LCD TV Models Competing in the 2026 TV Shootout (75-inch):

Although all six TVs were installed in the same room, along with four professional Sony broadcast monitors to use as reference displays, the competition was actually split into two groups: OLED on the left (all in 65-inch sizes) and RGB-lit LED/LCD TVs on the right (all in 75-inch sizes). Unlike in previous years, the judges were split into two groups as well: one group to judge the OLED TVs and one group to judge the RGB LCD TVs.

OLED vs. RGB Backlit LCD TV

OLED (Organic Light Emitting Diode) TVs use self-illuminating pixels to create the on-screen image. This allows them to create perfect black levels and outstanding contrast as each individual pixel and be turned fully on or fully off, or anywhere in between. If a star in deep space lights up exactly one pixel and is surrounded by pitch black space, that’s not a problem for an OLED TV.

LG’s G6 OLED TV, the company’s flagship model, uses a quad-stack Tandem RGB OLED (also known as “W-OLED” (White OLED) because it uses a white OLED element. But each white OLED uses red, green, white and blue sub-pixels for color reproduction. Meanwhile Samsung’s S95H and Sony’s BRAVIA 8 II OLED TVs use QD-OLED (Quantum Dot Organic Light Emitting Diode) technology in order to create the image. QD-OLED uses a blue OLED pixel combined with quantum dot technology to create the on-screen image and colors. While both technologies are capable of comparable brightness and contrast, QD-OLED TVs typically have slightly wider color gamut, which means they can reproduce a wider range of colors than W-OLED TVs.

RGB backlit TVs, sometimes known as “Micro RGB” TVs, differ from traditional LCD TVs in that they use multi-color red, green and blue LED backlight elements instead of a single color blue or white backlight. This allows RGB-lit TVs to offer higher efficiency, higher brightness and wider color gamut reproduction. But because RGB lit LCD TVs are still actually LCD TVs, there is no one-to-one match from lighting zone to each pixel. One RGB backline “zone” may serve hundreds or even thousands of individual pixels, so RGB-lit LCD TVs can have issues with uniformity, elevated black levels, blooming and haloing. When viewing a single-pixel star against a black space background, you will normally see a faint glow around the lit up pixel, due to the backlight bleeding through the LCD shutters. And that’s very difficult (if not impossible) to completely fix. But you can make it better by including a large number of backlight zones and use local dimming to control how these lighting zones operate.

All of the RGB-lit LCD TVs in this year’s Shootout include advanced local dimming, with a high number of dimming zones, so many of the traditional drawbacks of LCD TVs are minimized. We also noticed that all three of this year’s RGB-lit TVs offered off-axis viewing that was superior to most other LCD TVs, though still not as good as OLED TVs.

Managing Ambient Lighting

In terms of bright room viewing, four of the TVs in this year’s competition include some form of advanced glare reduction to minimize the reflection of in-room light fixtures or sunlight. The Samsung S95H and R95H use Samsung’s “Glare Free” screen coating, now in its third generation. The Sony BRAVIA 9 II uses Sony’s new “Black Screen Immersive Pro” anti-glare coating. The LG G6 OLED TV uses LG’s new “Reflection Free Premium” coating. All of these are effective at minimizing the direct reflection of ambient light, but they can impact perceived black levels.

Originally, each TV was going to be judged both with its best “out of the box” default video mode selected as well as in bright rooms, but some technical glitches with the peripheral equipment caused delays so those two parts of the testing were omitted. All of the evaluations that we did were done with professionally calibrated TVs in a light-controlled dark room setting. But having seen and experienced all three flavors of glare reduction, we feel that Sony’s Black Screen Immersive Pro coating is most effective at minimizing the reflection of ambient light without excessively washing out colors or changing the hue of black backgrounds.

Judges Assemble

As in previous competitions, all TVs were evaluated by professional judges from within the A/V community: tech journalists (including yours truly), content creators, trained calibrators and creative video content professionals. The pool of ten judges was split with five judging the OLED TVs and five judging the RGB-lit LED LCD TVs.

In addition to the TVs themselves, which were all professionally calibrated by expert ISF calibrator DeWayne “D-Nice” Davis, there were four professional Sony broadcast monitors on hand (BVM-HX3110), which typically sell for around $40,000 each. These are the pro monitors used by studios and mastering houses worldwide when grading and mastering movies and TV shows for home use. Judges were instructed to compare the image on the TVs with those on the broadcast monitors and rank each TV from 1 (not good) to 5 (best) based on how close they came to the reference monitors.

The evaluation featured video test clips, stills and real world movie scenes chosen to highlight specific areas of video performance such as contrast, brightness, color volume, color accuracy, motion reproduction, HDR tone mapping and image uniformity. After the tests, the judges’ scorecards were tabulated in order to crown a new “King of TV” for 2026. To be precise, two kings were crowned: one for OLED and one for RGB-lit LCD TVs.

As they have since the very first TV Shootout in 2004, consumers can use the results of the event and the judges’ scorecards to select the best TV for their own liking and specific use-case. Performance in both bright and light-controlled rooms will be considered.

As in previous years, it “takes a village” to put the event together, with many folks working behind the scenes to prepare for and run the event. It also takes a fair amount of gear. AVPro Global supplied the latest state-of-the-art switching, distribution, and test equipment for the event. Magnetar’s UDP-900MKII flagship 4K universal disc player and Kaleidescape’s Strato K 8K-capable Media Player were the reference sources for content with a Panasonic UHD Blu-ray Player also stepping in when the Magnetar player started misbehaving.

And the Winners Are…

Since the event was effectively separated into two parallel contests, with one set of judges rating the OLED TVs and another set of judges rating the RGB/LCD TVs, two winners were declared for 2026, a “King of OLED TV” and a “King of Micro RGB TV.” The winner in the OLED category, by a pretty healthy margin was the Samsung S95H QD-OLED TV. The winner in the Micro RGB category was the Sony BRAVIA 9 II True RGB TV. And even though they were effectively two different contests, the Samsung S95H had such higher scores overall that it was declared the overall winner of the 2026 TV Shootout.

The final detailed scorecards from the event have not yet been published, but when they are, we will be sure to update this article.

The Bottom Line

Having spent a significant amount of time with both Sony TVs and both Samsung TVs, and also having more limited hands-on time with both LG models, I was not surprised by the results. OLED TVs still have significant advantages over LCD TVs, even advanced RGB-backlit LCD TVs, particularly when it comes to picture uniformity, black levels, contrast and color saturation. I rated the Samsung S95H as the winner in my own judging, but I would personally take any of the OLED TVs over all of the Micro RGB TVs as my personal television.

The LG OLED did have some issues with color reproduction, with cooler whites and slightly muted colors, but these were only noticeable when compared to the reference monitors and to the two other OLED competitors. At home with a single TV on your wall or credenza, you’ll definitely appreciate the OLED TV’s deep inky blacks and outstanding contrast. And even though the Sony BRAVIA 8 II QD-OLED TV won last year’s TV Shootout (and its predecessors won for the prior three years), that TV is starting to show its age in processing, compared to the brand new Samsung S95H and LG G6. In my opinion, the LG G6 had stronger overall performance than the BRAVIA 8 II OLED, and any of the RGB-lit LCD TVs, but the Samsung S95H dominated the event due to its strong overall performance.

But does this mean everyone should go out and buy a Samsung S95H? Not necessarily. Some potential buyers are turned off by the S95H design, which features a large floating metal frame around the outside out the TV (which is not removeable). In my opinion, it makes the TV look more like artwork, particularly when it’s displaying art from the Samsung Art Store, but others prefer a more subtle narrow bezel. Also, the Samsung “Glare Free” matte screen, while extremely effective at reducing reflected light, can raise the black levels and add a subtle hint of purple to those blacks when viewed in a bright room. But if this bothers you, you can always just turn off the lights or draw the curtains.

Meanwhile, for those who are looking for a very large screen TV (above 83 inches), OLED is not really a viable option. Micro RGB TVs like the ones that competed in this year’s TV Shootout are available in sizes up to 115 inches diagonal. With the shift to RGB backlighting, LCD TV continues to improve in picture quality and the benefits of wider color gamut reproduction and higher brightness make them a solid choice if size, bright room viewing and picture performance are important to you.

While there is sure to be some controversy with the results (no one likes when their personal favorite TV doesn’t win), the TV Shootout remains one of the premier TV comparison events as it pits the top performing TVs against each other and hits them with a wide range of content to reveal each of their strengths and weaknesses. The event’s results should help inform TV buyers’ choices through the remainder of 2026 and beyond.

Where to Buy the Winning TVs?

For more information, all TVs are available at Value Electronics, Best Buy, and Crutchfield.

And special thanks to Robert and Wendy Zohn at Value Electronics for hosting the event.

Related Reading:

In brief: Remember the teenage GTA 6 hacker who was sentenced to life in a secure hospital facility in 2023? The former Lapsus$ member has been released from the unit and is now in prison awaiting a retrial for his crimes.

Back in September 2022, one of the biggest leaks of all time hit the internet when over 90 videos and screenshots of the still-unreleased GTA 6 were posted to the GTAForums. They originated from a then-18-year-old hacker who also claimed to be behind an Uber breach that occurred a few days earlier.

A couple of days after the leak, the City of London Police arrested the hacker, Arion Kurtaj, thanks to assistance from the FBI and NCCU. He was accused of causing almost $3 million of damage to Uber and accessing around 5,000 Revolut customers’ information. Kurtaj and a 17-year-old were also accused of blackmailing BT Group (Britain’s biggest broadband provider) and mobile operator EE between July and November 2021; they demanded a $4 million ransom. The pair are also alleged to be behind a hack on Nvidia.

In July 2023, Kurtaj was assessed by psychiatrists as unfit to stand trial due to his “acute autism.” That meant a jury could only determine whether he committed the crimes he was accused of, rather than finding him guilty or not guilty.

A mental health assessment also found that he continued to express an intent to return to cybercrime as soon as possible, adding that “he is highly motivated” to engage in more criminal activities.

sentenced in December 2023 the court heard that he will remain at a secure hospital for life unless doctors deem him no longer a danger. But he is now out and will face a conventional criminal trial in November. I have been trying to get this information confirmed with police for weeks after…

– Joe Tidy BBC News (@joetidy.bsky.social) 14 July 2026 at 15:38

Even being under police supervision didn’t stop the hacker. While staying in a hotel under police protection and despite having his laptop confiscated, Kurtaj was able to hack Rockstar using an Amazon Fire TV Stick provided in the room and a newly purchased phone, keyboard, and mouse. He broke into Rockstar’s internal Slack system and threatened to release the source code if the company did not contact him on Telegram.

Kurtaj was ordered to remain in the secure hospital indefinitely, unless doctors determined that he was no longer a danger. BBC correspondent Joe Tidy says he has now been moved out of that facility and into a regular prison awaiting retrial in November. Ironically, it’s the same month GTA 6 finally launches on consoles.

Tidy said it’s unclear why Kurtaj has now been deemed fit to face a conventional criminal trial after spending more than two years of being in a secure hospital.