Fewer bad actors are targeting Google Play with malicious apps, the company says, a shift that the tech giant credits with its increased investments in proactive security systems and AI technology.

In its latest Android app ecosystem safety report released on Thursday, Google said it prevented 1.75 million policy-violating apps from being published on Google Play in 2025, down from 2.36 million in 2024 and 2.28 million in 2023.

The annual report offers a look at how Google is keeping Android users safe by reviewing and monitoring apps to protect against malware, financial fraud, privacy invasions, sneaky subscriptions, and other threats.

For instance, Google says it banned more than 80,000 developer accounts in 2025 that had tried to publish these types of bad apps. That figure is also down year-over-year from 158,000 in 2024, and 333,000 in 2023.

Google touted how its investments in AI and other real-time defenses have helped fight these sorts of threats, but also how they served as a deterrent.

“Initiatives like developer verification, mandatory pre-review checks, and testing requirements have raised the bar for the Google Play ecosystem, significantly reducing the paths for bad actors to enter,” the company’s blog post explained, adding that its “AI-powered, multi-layer protections” have been “discouraging bad actors from publishing malicious apps.”

Google noted it now runs over 10,000 safety checks on every app it publishes and continues to recheck apps after publication. The company has also integrated its latest generative AI models into the app review process, which has helped human reviewers find more complex malicious patterns faster. Google said it plans to increase its AI investments in 2026 to stay ahead of emerging threats.

In addition, Google said it prevented more than 255,000 apps from gaining excessive access to sensitive user data, a figure that’s down from 1.3 million in 2024. The company also blocked 160 million spam ratings and reviews last year, and prevented an average 0.5-star rating drop for apps targeted by review bombing.

Meanwhile, Android’s defense system, known as Google Play Protect, identified more than 27 million new malicious apps, and warned users or blocked the app from running. That’s an increase from the 13 million non-Play Store apps identified in 2024 and five million seen in 2023. These increases seem to suggest that bad actors are now more often avoiding the Play Store when targeting users with their malicious apps.

When the Executive Branch soft-launched in Washington, DC, last spring, the private club’s initial buzz centered on its starry roster of backers and founding members. The president’s eldest son, Donald Trump Jr., is one of the club’s several co-owners, according to previous reporting. Founding members reportedly include Trump administration AI czar David Sacks and his All-In podcast cohost Chamath Palihapitiya, as well as crypto bigwigs Tyler and Cameron Winklevoss.

“We wanted to create something new, hipper, and Trump-aligned,” Sacks said at the time. Proximity to Trumpworld didn’t come cheap; though the club headquarters is located in a basement space behind a shopping complex, fees to join are reportedly as high as $500,000.

The initial wave of press for the MAGA hot spot identified Trump Jr. and his business associates Omeed Malik, Chris Buskirk, and Zach and Alex Witkoff as the club’s co-owners. A Mother Jones report later revealed the involvement of David Sacks’ frequent business associate Glenn Gilmore, a San Francisco Bay Area real estate developer who is given a variety of titles on official documents, including co-owner, managing member, director, and president.

But according to corporate filings reviewed by WIRED, there’s another key figure whose involvement has not been previously reported and whose connection to its more famous founders remains unclear: Sean LoJacono, a former Metropolitan Police Department cop in Washington, DC, who gained local notoriety for his role in a stop and frisk that resulted in a lawsuit.

According to the legal complaint, in 2017, after questioning a man named M.B. Cottingham for a suspected open-container-law violation, LoJacono conducted a body search. A recording of the incident went viral on YouTube, sparking intense debate over aggressive policing tactics. “He stuck his finger in my crack,” Cottingham says in the video. “Stop fingering me, though, bro.” The next year, the American Civil Liberties Union of the District of Columbia sued LoJacono on behalf of Cottingham, alleging that LoJacono had “jammed his fingers between Mr. Cottingham’s buttocks and grabbed his genitals.” Cottingham agreed to settle his lawsuit with LoJacono and was paid an undisclosed amount by the District of Columbia (which admitted no wrongdoing) in 2018.

The MPD announced its intention to dismiss LoJacono following an internal affairs investigation, which concluded that the Cottingham search was not a fireable offense but that another search he had conducted the same day was. By early 2019, LoJacono had appealed his dismissal, arguing in well-publicized hearings that he had conducted searches according to how he had been taught by fellow officers in the field. Initially, the dismissal was upheld. However, the police union’s collective bargaining agreement enabled LoJacono to further appeal to a third-party arbitrator, which in November 2023 ruled in LoJacono’s favor.

Instead of returning to the police force, though, LoJacono has gone down a different path. A LinkedIn account featuring LoJacono’s name, likeness, and employment history lists his profession as “Director of Security and Facilities Management” at an unnamed private club in Washington, DC, from June 2025 to the present. Official incorporation paperwork for the Executive Branch Limited Liability Company filed to the Government of the District of Columbia’s corporations division in March 2025, shortly before the club launched, lists LoJacono as the “beneficial owner” of the business. The address listed on the paperwork matches the Executive Branch’s location. Donald Trump Jr. and other reported owners are not listed on the paperwork; Gilmore is listed on this document as the company’s “organizer.”

The paperwork indicates that LoJacono is considered a beneficial owner of a legal entity associated with the Executive Branch. But what does that mean, exactly?

David Silver, a British AI researcher known for his role at Google’s DeepMind lab, has helped build some of the most influential AI systems and is now leading his own ambitious start-up. He is in the middle of raising a $1 billion seed round for his new London-based venture, Ineffable Intelligence.

If the fundraising will be completed, would be the largest seed round ever seen in Europe.

The round is being led by Sequoia Capital, with titans such as Nvidia, Google and Microsoft reportedly in talks to participate, according to industry sources familiar with the deal.

The proposed investment would place Ineffable Intelligence’s pre-money valuation at around $4 billion, an eye-watering figure for a company that hasn’t yet shipped a product.

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!

David Silver’s name is synonymous with some of the most memorable milestones in AI. During more than a decade at Google DeepMind, he helped develop AlphaGo, the first AI to defeat a world champion at the ancient game of Go, and later AlphaStar, which bested professional players in StarCraft II.

His work has shaped how modern AI systems learn and make decisions.

In late 2025, David Silver left DeepMind, where he also contributed to the development of the Gemini family of large language models, to launch Ineffable Intelligence. .

In the current days billion-dollar seed rounds are no longer unheard of, but they remain exceptionally rare, especially in Europe. 

If Sequoia and its partners follow through on their commitments, the round would serve as both a vote of confidence in David Silver’s vision and a broader signal that venture capitalists are increasingly willing to fund ambitious, research-driven AI ventures well before traditional product milestone

This could be a new chapter for European AI

The sheer size of the potential round also reshapes the narrative around Europe’s role in the global AI race.

Historically, most of the industry’s capital and rich technical talent have gravitated to Silicon Valley and other U.S. hubs. A £737 million-plus round led by a top U.S. investor shows that world-class AI ideas can still attract world-class capital in Europe.

David Silver’s dual role – as an entrepreneur and as a professor at University College London,  hints at a blend of academic ambition and commercial pressure that may define the next era of AI innovation.

His thesis suggests that the path to genuinely powerful AI may not be simply scaling up models, but teaching systems to learn and improve through real-world experience, a direction he believes could outpace approaches grounded solely in data accumulated from human interactions.

Whether Ineffable Intelligence ultimately delivers on its lofty vision remains to be seen.

But the enthusiasm around its fundraising reflects a broader trend: AI is entering a phase where conviction in founders and ideas can translate directly into unprecedented capital flows.

For Europe, a successful $1 billion cap table would be a milestone moment, not just financially, but symbolically,  in positioning the continent as a contender for the future of AI research and deployment.

Google’s announcement that its Gemini app now writes music for you isn’t just one of those “blowing my mind” product updates. It feels like a symbolic surrender to a long-standing refrain from Big Tech: creative work is now just another checkbox for a machine. 

If you don’t know what I am talking about, yesterday Google launched a new feature, Lyria 3, in the Gemini app, that allows us to cook up 30-second tracks complete with lyrics and cover art from a text prompt or a photo, of course, generated by Nano Banana; basically, no instruments, no experience, no pesky tactile skill required.

It’s essentially a LEGO set for “songs” that lasts about as long as a TikTok loop. They say it’s designed for YouTube Creators, and I tend to agree with them, because you can’t make to much with 30 seconds. 

Still, the underlying problem is another one, as I am seeing different projects/songs made with AI, including AI artists. And this is what I want to highlight in this piece.

“Behind every beautiful thing, there’s some kind of pain, “ said Bob Dylan, and I could not agree more.

If we take a look at the history ( art, music, literature, poetry, and so on), the main fuel for creation was indeed pain. 

Now, how should I put this? Probably the only pain that Lyria can feel is more like a faint server-overload alert than heartbreak.

Real songwriters know that soul isn’t born in a 30-second prompt, it’s extracted through years of mistakes, late nights, losses, and tiny revelations.

Call it a toy if you like. Google will, too. 

They even watermark the outputs with a SynthID tag so the 30-second ditties are officially AI-generated, not “inspired.” That’s a nod to copyright concerns, but it also reads like an admission: these aren’t really art, they’re chemical by-products of pattern statistics. 

What’s striking isn’t the novelty. Much of this has been possible in labs and APIs for years, and creators have been experimenting with generative music tools as collaborators. 

What Lyria 3 does, and what makes this moment worth watching, is normalising the idea that anyone can “write” a song with a chatbot and a mood descriptor. That’s not empowerment; it’s a devaluation of craft.

Just because you pay a subscription to Suno, that’s another AI music generator, and that one is more complex, that doesn’t make you an artist or a singer. Just because you learn how to write a prompt for any LLM model and generates you pages, you are not a writer. 

Imagine a world where every blog has AI-generated copy, not that we are not almost in the middle of this, and every company can churn out half-baked music for their ads or social posts.

In that economy, a professional songwriter’s unique skill becomes as optional as knowing how to use a metronome.

You could ask Gemini for an “emotional indie ballad about a lost sock,” and voilà, you have something. Whether it has actual coherence or soul is left to the listener to decide. It is fun to use it with your friends, shorts, to impress your date. 

Video: Gemini Lyria Music Generation Feature – Socks, uploaded by Google on YouTube

Still, Lyria 3’s music is capped at 30 seconds, and that’s no accident. It sidesteps deeper legal and ethical quarrels about training data and mimicry of existing works by keeping outputs short and legally fuzzy. That’s a thumbs up from me. 

But even within that limit, it’s now possible for someone with no craft or cultural context to generate riffs, lyrics, and chord progressions that sound, to the casual ear, adequately musical. In an attention economy obsessed with shareability, “adequate” quickly becomes plenty. 

This matters because real songs, the ones that endure, that carry human experience, aren’t just collections of musical atoms. They’re shaped by story, risk, cultural memory, and sometimes contradiction. 

One of my favorite artists, Tom Waits, said, ”I don’t have a formal background. I learned from listening to records, from talking to people, from hanging around record stores, and hanging around musicians and saying, “Hey, how did you do that? Do that again. Let me see how you did that.”

This was the research and prompting before, and it’s not only about reducing time, or getting things faster, and “having more time for you.”

It’s about the entire process, the contact with other artists, humans, and IDEAS. 

Those are qualities machines can mimic but not originate. When the machines own the first pass at creation, and the commercial ecosystem embraces that output because it is cheap and fast, the incentives shift. Not gradually. Suddenly.

The record industry is already grappling with AI. Streaming services, publishers, and even labels have begun experimenting with algorithmic playlists and automated composition.

What Gemini’s Lyria 3 does is extend that experiment to public perception. A whole generation may come to think that “making music” means typing a description and choosing a style. Songwriting becomes a UX problem, not a craft one.

That raises a serious question: in a world where AI can conjure up a half-decent hook on demand, what will distinguish professional artists?

If the answer is only brand story or marketing muscle, we aren’t celebrating creativity; we are monetising it out of existence.

Tech companies like Google will frame this as liberation. And in a literal sense, anyone who’s ever wanted to hear a short tune about a sock’s existential crisis now can. But liberation without value for the creator is just consumerism by another name. 

Lyria 3 might be good for GIF soundtracks and social clips, and TikTok viral reels, but it doesn’t make professional musicians obsolete; it makes their work less necessary to the platforms that reward hyper-consumable content.

That’s a different threat from outright replacement: it’s obsolescence by trivialisation.

If AI is going to be part of musical creation, then let it be as an assistant to the composer,  someone who improves ideas, not replaces them. What we’re seeing with Gemini is not collaboration but outsourcing.

And the lesson for artists isn’t to fear the algorithm. It’s to insist on clarity about where AI replaces labour and where it augments human sensibility.

Because once the marketplace equates the two, the humans who do the work will be the ones left asking for royalties in a language no one else wants to speak.

And, as a personal recommendation, not sponsored, there are streaming platforms like Deezer that have built AI detection tools that flag and label AI-generated tracks, excluding them from recommendations and royalties so that human songwriters aren’t buried under synthetic spam, and consumers can make the difference between AI and human. 

If you care about preserving real artistry in a world of text-to-tune generative models, start paying attention to how platforms handle AI-tagging and choose services that give you transparency about what you’re actually listening to.

Yet, I’m not here to throw shade at Lyria 3; if anything, the idea of letting people turn a photo or a mood into a short track sounds like fun for casual use and creative experimentation. It is what Google says it’s meant for. 

Yet the reality is that as these models proliferate, we risk confusing novelty with art. And here, the big tech companies are not the ones to blame, but us. 

The race to advance conversational AI in the living room is heating up, with YouTube being the latest to expand its tool to smart TVs, gaming consoles, and streaming devices. 

This experimental feature, previously limited to mobile devices and the web, now brings conversational AI directly to the largest screen in the home, allowing users to ask questions about content without leaving the video they’re watching. 

According to YouTube’s support page, eligible users can click the “Ask” button on their TV screen to summon the AI assistant. The feature offers suggested questions based on the video, or users can use their remote’s microphone button to ask anything related to the video. For instance, they might ask about recipe ingredients or the background of a song’s lyrics, and receive instant answers without pausing or leaving the app. 

Currently, this feature is available to a select group of users over 18 and supports English, Hindi, Spanish, Portuguese, and Korean.

YouTube first launched this conversational AI tool in 2024 to help viewers explore content in greater depth. The expansion to TVs comes as more Americans now access YouTube through their television than ever before. A Nielsen report from April 2025 found that YouTube accounted for 12.4% of total television audience time, surpassing major platforms like Disney and Netflix.

Other companies are also making significant strides with their conversational AI technologies. Amazon rolled out Alexa+ on Fire TV devices, enabling users to engage in natural conversations and ask Alexa+ for tailored content recommendations, hunt for specific scenes in movies, or even ask questions about actors and filming locations.

Meanwhile, Roku has enhanced its AI voice assistant to handle open-ended questions about movies and shows, such as “What’s this movie about?” or “How scary is it?” Netflix is also testing its AI search experience. 

Another way YouTube has tried to improve its TV experience with AI is the recent launch of a feature that automatically enhances videos uploaded at lower resolutions to full HD.

Additionally, the company continues to launch other AI features, like a comments summarizer that helps viewers catch up on video discussions and an AI-driven search results carousel. In January, the company announced that creators will soon be able to make Shorts using AI-generated versions of their own likeness. 

Last week, YouTube launched a dedicated app for the Apple Vision Pro, too, letting users watch their favorite content on a theater-sized virtual screen in an immersive environment.

No matter how well your favorite streaming service’s algorithm knows you, come February, sometimes even the smartest technology can be swayed by the power of Valentine’s Day. Hence all those romance-heavy promos at the top of your screen, from Ryan Murphy’s Love Story to Netflix’s ever-proliferating Love Is Blind.

But love—romantic or otherwise—can be found in the oddest of places, including the radioactive wasteland of postapocalyptic Los Angeles, Westeros in the rare midst of relative peace, or behind the scenes of the latest MCU blockbuster.

Whether you’re in the mood for a reliable sci-fi gem or an enlightening new docuseries courtesy of Marty Supreme director Josh Safdie, February’s streaming lineup offers plenty of options to swoon over. Here are the 10 shows we’re falling for right now.

Star Trek: Starfleet Academy

Whirlpool has unveiled what it claims is an industry first: a refrigerator with a built-in nugget ice maker.

Announced ahead of KBIS 2026 in Florida, the new 36-inch Wide True Counter Depth French Door model integrates soft, chewable nugget ice directly into the door. This means no countertop machine is required.

For years, nugget ice has been the preserve of standalone makers cluttering up kitchen counters. However, Whirlpool’s approach folds it into a full-size fridge, complete with dual ice makers that dispense both traditional cubes and nugget ice from the same appliance.

It’s a small upgrade on paper, but for households that love the softer, chewable style popularised by cafés and fast-food chains, it removes the need for an extra gadget.

Photo credit: The New Camera
A new hands-on video has emerged, purportedly showing the DJI Osmo Pocket 4 in action, and it comes directly from a Malaysian store. The clip, provided by a local DJI outlet named DronesKaki in the Kuala Lumpur area, shows a customer messing with what appears to be a production unit.

The device retains the Pocket series’ signature compact dimensions. Its three-axis gimbal performs an excellent job of keeping the footage smooth, with no shakes visible even when moving about the store. The flip out screen on the bottom measures a few inches and is noticeably brighter than previously. You also have a joystick and a few buttons on the front, as well as a couple more concealed beneath the screen that we’re not sure what they do.

Sale

DJI Osmo Pocket 3, Vlogging Cameras with 1” CMOS & 4K/120fps Vlog Camera, 3-Axis Stabilization, Fast…

• Capture Stunning Footage – This vlogging camera features a 1-inch CMOS sensor and records in 4K resolution at an impressive 120fps. Capture…
• Effortlessly Frame Your Shots – Get the ideal composition with Osmo Pocket 3’s expansive 2-inch touch screen that rotates for both horizontal and…
• Ultra-Steady Footage – Say goodbye to shaky videos. Osmo Pocket 3’s advanced 3-axis mechanical stabilization delivers superb stability. Enjoy smooth…

One visible new feature is the camera’s tiny built-in LED light, which appears to be mounted on an adjustable arm. That will come in handy when shooting in low light, and you won’t have to carry around any extra equipment. You may also spin it to face a different direction to give yourself more angle options.

The Pocket 4’s rumored specs sound promising, with a beefed-up 1 inch CMOS sensor that should help with low-light work and the ability to record 4K video at 120fps. According to speculations, you may be able to get 4K at 240fps slow motion or 6K at 30fps. Autofocus is said to be more faster now, and AI-assisted tracking helps keep your subject in frame.

The battery reportedly lasts up to 200 minutes, which is significantly longer than the Pocket 3’s 166 minutes. The device still weighs 179 grams and now includes Wi-Fi 6, which should make file transfers much faster.

According to the video, there is a SuperPhoto mode that utilizes artificial intelligence to analyze your environment and perform smart processing. That should provide you with some decent stills, up to 33MP at least. This might be a feature you’ve seen on other action cameras, but it’s a welcome addition to the DJI ecosystem.

With the leak, prior FCC records suggest a March 2026 arrival date. Some anticipate a late February or early March release, and while there are rumors of a Pro version with additional features such as dual cameras, the video we viewed focuses entirely on the standard model.
[Source]