Amazon Echo Show 8 (4th Gen) for $180: This is a solid smart speaker with Amazon’s latest physical design and Alexa+ right out of the box, but it’s not a sound improvement over the older model, so I’d personally pick the third-gen option (see above) while it’s still available or upgrade to the Echo Show 11 ($220).

Amazon Echo Show 15 for $300: The Show 15 exists somewhere on the continuum of being a smart display and a smart TV, but it doesn’t quite fully nail being either. The widgets are fun to use since you can add so many to the Show’s 15-inch screen, but I’ve tried this device a few times, and I’ve always walked away underwhelmed. The Show 15 has grown on me while using it with Alexa+, though, particularly with a stand ($125) to sit on my desk. But it’s still larger than I need for day-to-day tasks, but smaller than what I’d want from a television.

Apple HomePod for $299: Apple’s flagship smart speaker has a muddy midrange and high-end, which is disappointing for the price point. The HomePod does have a lot of bass, though, if that’s your jam. If you want an Apple-powered smart speaker in your home, the Mini is a third of the price and has nearly identical capabilities to the full-size model.

Bang & Olufsen Beosound Level for $2,250: This is a gorgeous—though seriously expensive—speaker that’s built to last. The company has designed the high-end model to be repairable and upgradable over time. Made of natural fabric and wood, it’s a high-design flat speaker that comes with Google Assistant onboard—or you can buy it without a smart assistant for the same price.

JBL Authentics 200 for $200: This was my previous pick for a third-party smart speaker, but I’m uncertain of its access to Amazon and Google’s newest assistants. I’ll retest it once I confirm if it will gain access to one (if not both) assistants.

Sonos Era 100 for $189: Another third-party option, but it won’t grant access to Google support. You can connect it to Amazon Alexa, though. Plus, Sonos is a great investment if you’re really looking for a great speaker that can have smarts—but its smart assistant isn’t the primary feature.

Sonos Beam Gen 2 for $369: This is an older version of the Sonos Beam that still has Google support, but I’m uncertain if it’ll gain access to Gemini for Home.

WiiM A10 for $229: This speaker doesn’t have a voice assistant, but it does have compatibility with AirPlay 2, Google Cast, and Siri to be used as a Bluetooth speaker. WIRED reviewer Parker Hall says it reminds him of a Sonos speaker, but that it can instantly connect with Spotify Connect—faster than any other speaker he’s tried.

FAQs

Power up with unlimited access to WIRED. Get best-in-class reporting and exclusive subscriber content that’s too important to ignore. Subscribe Today.

Among those who turned down job offers, nearly a third did so because it didn’t pay enough

If you’ve just graduated and are stressed about your starting salary, you’re not alone.

According to a CNA report, a Ministry of Manpower (MOM) survey of residents aged 22 to 28 found that graduates across most disciplines are earning less than they expected when entering the workforce.

And with global uncertainty still looming, analysts are telling fresh grads to manage their expectations—or risk sitting on the job hunt longer than they’d like.

How big is the salary gap?

CNA cited figures from MOM to show the gap between what fresh grads expected to earn and what they actually took home.

IT graduates were among those with the highest expectations at S$6,000, but averaged S$5,150—a $850 shortfall. Engineering sciences grads expected S$5,000 and landed at S$4,450.

The disparity was even larger among business administration graduates, who expected S$5,000 but earned approximately S$4,000. Graduates in natural and mathematical sciences also saw one of the widest gaps, with expected salaries of S$5,000 compared to actual earnings of S$3,700.

Only a handful of disciplines met or exceeded expectations.

Law graduates earned an average of S$7,500 despite expecting S$6,500, while education graduates earned S$4,000 against expectations of S$3,800. Fine and applied arts graduates reported earnings that matched expectations at S$3,500.

Low salaries remain the top reason for offer rejections

MOM’s survey also shed light on why some graduates turn down job offers.

Among university graduates who rejected an offer, nearly one-third (30.6%) cited low pay as the main reason.

Another 26.7% said they were holding out for better opportunities, while 11.3% were uninterested in the role itself. Around 10% pointed to unsuitable workplace environments, and 6.1% said the jobs lacked clear career progression.

Professor Lawrence Loh from the NUS Business School told CNA that graduates’ emphasis on salary reflects Singapore’s rising cost of living.

He noted that many young professionals want to secure the highest possible starting salary because future pay increments can be difficult to obtain. A stronger starting point, he said, provides better long-term earning potential and career mobility.

While MOM expects wages to continue rising, the ministry noted that employers are taking a more cautious approach to salary increases due to global economic uncertainty and inflationary pressures.

Hiring conditions are becoming more challenging

That said, analysts also told CNA that graduates may have to “taper down their expectations” or risk delaying their entry into the workforce.

Anurag Garg, country lead at recruitment firm Michael Page, said prolonged job hunts could leave candidates frustrated and potentially cause them to miss out on suitable opportunities.

On the employer side, companies competing for top talent may encounter higher rates of offer rejections, which can lengthen hiring timelines.

Garg added that mismatched expectations could also contribute to underemployment, where workers take on roles that do not fully utilise their skills and qualifications.

As hiring conditions become more challenging, analysts say both employers and graduates will need to strike a balance between compensation expectations and market realities.

• Read other articles we’ve written on Singapore’s job landscape here.

Featured Image Credit: The Recruiter

When the 29-year-old Samuel Rizzon is asked what he does, he answers with a single word: “developer.” While accurate, it’s a modest label for someone whose work has stretched well beyond writing code.

At an age when many engineers are still settling into a single specialty, Rizzon has built products embraced by large enterprises, online classrooms, and the open-source community, three arenas that rarely reward the same instincts. His is a story of versatility, of an engineer who has never been willing to be only one thing.

From a bedroom app to a billion documents

Interested in technology and building software from a young age, Rizzon developed and shipped his first product at 19: a Bible quiz he published to the Play Store and the App Store in 2015. It picked up 22,000 downloads, and that response was enough to convince him that making things people actually used was worth pursuing. Not long after, he joined TOTVS, Brazil’s largest technology company, where he would spend the next five years and lay the foundation of his career.

That foundation took shape around a single product. It started as a proof of concept for one client that wanted a way to sign documents digitally. Rizzon wrote it from scratch, and the prototype worked well enough to become a product in its own right. It grew into a standalone electronic signature platform comparable to DocuSign, and today it processes more than a billion documents for over a million customers.

Building it was largely a solo effort. Working before AI coding assistants existed, Rizzon architected the whole stack himself, from an Angular front end and a C# back end to a Chrome extension and a desktop application that reverse-engineered the physical A1 and A3 devices Brazilians use to authenticate documents. As the product matured, a team formed around him, eventually reaching roughly 10 engineers, designers, and product staff, with Rizzon leading the work that turned the prototype into a full product line.

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!

From there he spent a year at the consultancy CI&T before taking a remote role as a full-stack engineer for a New York startup, a job that gave him his first direct contact with the U.S. technology scene.

Gaining experience as a founder

Around the same time, Rizzon set out to build a company of his own. He ran it out of his room in Brazil, without investors, without a team, and without a network to lean on. What he had was persistence, and it showed: he took the business from nothing to 30 paying customers across Brazil, the United States, and Ireland, with 8,000 people using its web app.

Since there was no one else to do it, he handled sales, client conversations, support, and marketing himself, the parts of a business most engineers never touch. He even started a YouTube channel during this period, which grew to 3,000 subscribers.

He doesn’t romanticize how hard it was, and he is especially frank about the difficulty of doing it from Brazil, far from any real startup network. “I had nothing, really nothing,” he points out. “It was just me in my room. Creating something and trying to sell it and reach customers. It was a very specific niche, and it was a hard niche.” That isolation forced him to operate on his own, and the founder instincts it produced would resurface later in the viral consumer projects that made his name.

A one-click fix that found 150,000 users

Amidst the 2020 COVID-19 pandemic, with so much of work and school suddenly happening over video, Rizzon built a Chrome extension that muted every participant on a Google Meet with a single click, a product that could solve a problem he kept running into himself. The fix was simple, but it turned out plenty of other people had the same complaint.

That became clear fast. Within a year the extension had reached 150,000 users, almost all of them arriving by word of mouth. Its most devoted users were teachers, who were running online classes for 15 to 30 students and had no way to quiet the room without clicking each child one by one. “It was a pain for me, and I just fixed that with an extension,” Rizzon says. “It ended up being useful for a lot of teachers in particular.”

The traction caught the attention of the founder of MP3.com, who emailed Rizzon with an offer to buy it. He sold, marking his first exit and an early sign of the instinct for shipping consumer products that would shape his later work. He has stayed close to open source since, serving as co-founder and core component developer of Zard UI, a shadcn-style component library for Angular developers that has crossed 1,000 stars on GitHub.

The city he engineered to go viral

After years of shipping one project after another, the one that finally broke through was GitCity. The idea came from a post on X about rendering a city, and Rizzon had a first version live within a day. He didn’t write any of it by hand; instead, he built the entire codebase with Claude Code. What it produced was a pixel-art 3D metropolis that renders GitHub developers as buildings, with one structure per coder.

“On the first day, when I had the idea of creating the city, I noticed that this could be a viral product,” he says. “So I prepared and made everything to go viral.”

People took to it immediately. In its first week the city grew from 12,000 buildings to 40,000, and it currently holds more than 80,000. Over two months GitCity drew 180,000 visitors, more than five million social media views, and 5,000 GitHub stars, with roughly 20 people contributing code. Rizzon’s own audience grew alongside it, climbing from 200 Instagram followers to 6,000 and an X account to nearly 4,000.

None of that happened by chance. Rizzon treated distribution as part of the product itself, wiring a one-click “share on X” button into every action a user could take. He also added a feature that lets one building attack another, which fires off an email to the target and pulls them back in to retaliate, and he also opened the experience with a cinematic shot of the skyline and made the 3D rendering run smoothly on phones.

Inspired in part by the indie developer Pieter Levels, he’s also begun earning money from it, taking in $2,000 from sponsored buildings and lining up companies to back a week-long event in which users hunt down a “dark boss” hidden in the city.

The project did more than rack up numbers; it brought recruiters from Delphi, who were looking for someone to carry that same obsession with user experience into their consumer product, and he’s now joining as a product engineer on their San Francisco team. He treats the move less as a destination than as a long apprenticeship, a chance to build a network and learn how the U.S. startup world actually works before starting a company of his own.

More than a developer

Whether the right title is developer, founder, or product engineer, Samuel Rizzon has spent a decade declining to choose just one. The same engineer who built a signature platform now handling more than a billion documents at TOTVS also turned GitCity into a viral calling card, proof that the instinct to ship and the obsession with how a product feels follow him regardless of the label.

When I first opened the box the Oura Ring 5 comes in, my first thought was, “Wow, that’s tiny.” My second thought was that this is a smart ring a lot of people have been waiting for. 

As someone who got quite used to wearing the Oura Ring 4 Ceramic, I was surprised how noticeably smaller and lighter the new Ring 5 is.

The Ring 5, which Oura describes as the world’s smallest smart ring, is 40% smaller than its predecessor, measuring 6.09 mm wide compared to the Ring 4’s 7.90 mm, and 2.28 mm thick compared to the Ring 4’s 2.88 mm. Although the exact weight depends on your ring size, the Ring 5 weighs between 2 grams and 2.69 grams, while the Ring 4 weighs between 3.3 grams and 5.2 grams. 

I found that these changes dramatically improve the ring’s comfort, and also make it more aesthetically pleasing compared to its predecessor. The ring no longer screams smart ring, and blends in with the rest of your jewelry. Oura says the Ring 5 is designed to look and feel like any other ring, and I think the company has achieved that. 

The Ring 5 starts at $399.

Whenever I saw people discussing Oura’s smart rings, there always seemed to be two opposing viewpoints. One side swore the ring had changed their life, and the other argued that it was too bulky and that they would never consider getting one. I think the Oura Ring 5 changes the game and appeals to a larger audience, including those who shied away from smart rings due to their bulkiness. 

Oura was aware of the demand for a smaller ring. The company told me that users had been asking for a thinner and more compact design, prompting the company to comply. Of course, Oura has also had to update its rings in response to competition from subscription-free rivals like RingConn and Ultrahuman, both of which sell rings lighter than the Ring 4. 

While I never found the Oura Ring 4 overtly uncomfortable, the Ring 5 feels noticeably better on my finger. With the Ring 4, I was always aware that I was wearing it, but with this latest model, I often forget it’s there, which is great for people like me who don’t always wear jewelry. 

I also found that the ring’s smaller size makes it more comfortable to wear at night for tracking sleep and health metrics. A smart ring is more comfortable than a smartwatch for nighttime wear, and the Ring 5’s smaller design lets it be even less noticeable at night.

As for battery life, the Ring 5 lasts between six and nine days, compared to the five to eight day range on the Ring 4. In my experience, the improvement seems to hold up. The ring arrived 50% charged, and after about 30 minutes on the charger following the set-up process, it reached 75%. After five days of continuous wear, I still haven’t needed to charge it, and I’m down to about 25% battery. 

It’s worth mentioning that the Ring 5 comes in fewer sizes (sizes 6 to 13) than the Ring 4 (sizes 4 to 15). Oura told me that the fewer sizing options are due to the challenges of manufacturing smaller rings in the new form factor. The company said it chose to focus on the most popular sizes, and is monitoring demand for the discontinued sizes.

The Ring 5 comes in six finishes, including a redesigned Gold with a truer gold tone, an updated Deep Rose with a copper-like look, plus Silver, Brushed Silver, Black, and Stealth.

Oura sent me the Gold version, and I really like how it looks. Unlike previous gold Oura rings, this new color doesn’t have a yellow tinge to it, and instead has a subtler tone that feels closer to actual gold jewelry. It’s worth noting that the Gold finish, along with the Stealth and Deep Rose finishes, is priced at $499, exactly $100 more than the standard finishes. 

As for durability, Oura says the Ring 5 is more scratch-resistant than previous generations thanks to a new finishing technique, but I can’t fully speak to durability yet considering I’ve only had it for five days.
There was a moment when I thought I had scratched the ring’s alignment guide line when I’d grabbed a rusty swing chain, but it turned out that rust had only rubbed on to the ring, and I was able to clean it off with a cloth.

Oura says the Ring 5’s new physical vapor deposition (PVD) coating process ensures the wearable retains its premium “out-of-the-box” look for longer. It will be interesting to see how this promise holds up.

The Oura Ring 5 is being launched alongside new software features that are also coming to the Oura Ring Gen3 and later products, including Blood Pressure Signals and Nighttime Breathing. I can’t speak to those yet, as they’re set to launch later this month. 

Overall, the Oura Ring 5 is a notable upgrade over the Ring 4 in terms of comfort and aesthetics, making it a great choice for anyone who has yet to buy a smart ring. As for people who already have the Oura Ring 4, the decision to upgrade depends on your budget and how much you value aesthetics, especially since the Ring 4 will get all of the new software updates.

Cambridge Audio’s Evo Series has been one of the British manufacturer’s real success stories over the past three years, and not because it tried to make separates disappear with marketing fog and a nice volume knob. The Evo 75, Evo 150, Evo CD, Evo S, and more recent Evo 150 SE have shown that Cambridge can marry serious hi-fi performance, clean industrial design, flexible connectivity, and pricing that is definitely not out of line with reality.

The new Cambridge Audio Evo 300, making its debut at HIGH END Vienna 2026, pushes that formula into far more muscular territory. Rated at 300 watts per channel from Hypex NCOREx Class D amplification in a dual-mono layout, the Evo 300 is being positioned as Cambridge Audio’s most powerful and sonically advanced streaming amplifier to date.

We have covered the Evo Series in depth, including the original Evo 150 and the newer Evo 150 SE, and the appeal has been obvious from the start: real hi-fi performance, a very solid streaming platform, flexible connectivity, and industrial design that does not look like it escaped from a server rack. The Evo 300 builds on that formula with more power, a balanced pre-amplifier stage, dedicated analog volume controls for each channel, and a switch-mode power supply designed to keep things stable when the amplifier is pushed hard.

300 Watts Per Channel From Hypex NCOREx

The headline number is hard to miss: 300 watts per channel into 8 ohms from a Hypex NCOREx Class D amplifier stage. That is a significant jump for the Evo platform and clearly aimed at listeners using larger rooms, more demanding loudspeakers, or both. Cambridge has not turned the Evo 300 into a traditional heavyweight integrated amplifier with a heat sink complex, but the move to NCOREx gives it a much stronger power story than the Evo 150 SE.

Cambridge has also gone with a dual mono layout, which matters because channel separation is not just a marketing term to make you feel better about your purchase. Better separation can help with stereo imaging, focus, and control, assuming the rest of the system is up to the task. The Evo 300 also uses a balanced preamplifier stage and dedicated analog volume controls for each channel, which Cambridge says are designed to preserve focus, dynamic range, and low level detail at different listening levels.

A switch mode power supply is part of the design as well, and the goal is consistency when the amplifier is pushed for longer periods. That will matter for anyone driving lower sensitivity loudspeakers or asking one box to do the work of a more ambitious separates system.

StreamMagic Gen 4 Handles the Network Side

The Evo 300 uses Cambridge Audio’s StreamMagic Gen 4 platform, which remains one of the stronger reasons to consider an Evo product in the first place. Streaming support includes Spotify Connect, TIDAL Connect, Qobuz Connect, Amazon Music, Deezer, UPnP, Internet Radio, and Roon Ready operation.

That gives the Evo 300 a lot of flexibility without forcing users into one ecosystem. Ethernet and WiFi are both supported, and Cambridge’s in-house platform means updates and feature improvements are not entirely dependent on a third party app that may vanish at some point.

Cambridge has taken some heat online over the past year over StreamMagic Gen 4 stability with specific music streaming apps, and those complaints should not be dismissed. A streaming amplifier lives or dies by the quality of its software, especially when buyers are spending real money on an all-in-one platform.

That said, the recent move to Fidelity Imports as Cambridge Audio’s U.S. distributor appears to have brought more urgency to the software side of the business, and the volume of complaints seems to be easing. As someone who has invested a significant chunk of change in Cambridge Audio products over the years, and who uses the StreamMagic app daily for both review purposes and personal listening, I find the platform quite good overall.

It is not perfect, but Cambridge has not sat on its hands. The company is clearly paying attention, and that matters if the Evo 300 is going to succeed as more than just a powerful amplifier with a nice screen.

The Evo 300 also supports Google Cast, Apple AirPlay 2, Google Home, Apple AirPlay multiroom, and Roon multiroom systems, which makes it easier to integrate into a wider home audio setup. For a product trying to replace multiple boxes, that matters.

ESS Sabre DAC With Serious Hi Res Support

Digital conversion is handled by the ESS Sabre ES9038Q2M DAC, with support for up to 32-bit/768kHz PCM and DSD512. Those numbers will make the hi-res crowd happy, although most listeners will spend the majority of their time with streaming services that do not get anywhere near those ceilings.

The more important point is that Cambridge is positioning the Evo 300 as a proper digital hub, not just an amplifier with a streamer bolted on because someone in marketing found a WiFi antenna. USB audio, optical Toslink, and coaxial digital inputs are included, giving users enough digital flexibility for a server, disc transport, TV, or legacy digital source.

HDMI eARC Makes It Living Room Friendly

The inclusion of HDMI eARC is one of the smarter practical decisions here. It allows the Evo 300 to sit in a living room system and handle TV audio without requiring a soundbar or AV receiver. This is still a two channel amplifier, so nobody should expect Dolby Atmos fireworks or rear channel theatrics, but for listeners who care more about music and still want better TV sound, HDMI eARC gives the Evo 300 a real advantage.

That also makes the Evo 300 easier to justify as the center of a daily use system. Music, TV, streaming, vinyl, headphones, and external sources can all run through one chassis. Fewer boxes. Fewer cables. Fewer reasons to mutter at the back of the rack with a flashlight in your mouth.

Vinyl and Analog Inputs Are Not an Afterthought

Cambridge did not ignore analog users. The Evo 300 includes a built-in moving magnet phono stage, RCA line input, and balanced XLR input. The MM phono stage will be enough for a lot of turntable owners, although moving coil cartridge users will still need an external phono preamp.

The balanced XLR input is also worth noting because it gives the Evo 300 more credibility as a serious integrated amplifier. Anyone using a higher end DAC, phono stage, or source component with balanced outputs can connect it directly rather than being forced through RCA only.

Subwoofer Control and Speaker Flexibility

The Evo 300 includes an adjustable subwoofer output with independent level and crossover control, plus optional high pass filtering for the main speakers. That is a very useful feature for most rooms, especially smaller spaces where full range loudspeakers can turn bass into soup before the first chorus arrives.

The optional high pass filter also matters because it can reduce the low frequency burden on the main loudspeakers. Used properly, that can improve system clarity and give the amplifier and speakers a little more breathing room.

Cambridge also includes dual speaker binding posts per channel, allowing users to create two separate speaker zones. That adds some installation flexibility, although buyers should pay attention to impedance and speaker matching before treating the Evo 300 like a nightclub distribution amp.

Headphones, Bluetooth, and Daily Use Features

A dedicated 6.3mm headphone output is included for personal listening, which is the right choice on a product at this level. The press materials do not position the Evo 300 as a dedicated headphone amplifier replacement, but the inclusion of a full size headphone jack keeps it useful for late night listening.

Wireless support includes Bluetooth 5.4 with aptX HD, which is fine for casual listening, but it is not the reason to buy the Evo 300. There is no Bluetooth LE Audio, aptX Lossless, or LDAC support, so the better path is StreamMagic, AirPlay 2, Google Cast, TIDAL Connect, Qobuz Connect, or Roon.

Bigger Display, Interchangeable Side Panels

The Evo design language carries over, but the Evo 300 gets a wider and more substantial chassis. It also features Cambridge Audio’s largest Evo display to date: a 7.8 inch color screen that can show album artwork, track details, system information, a clock, or VU meters.

The interchangeable side panels remain part of the package, giving owners a way to better match the amplifier to their room. That may sound cosmetic, but Cambridge has understood something many hi-fi brands still treat as classified information: products that live in shared spaces need to look good from more than three feet away.

The Bottom Line

The Cambridge Audio Evo 300 is not just an Evo 150 SE with a bigger number on the box. The real news is the move to 300 watts per channel into 8 ohms from Hypex NCOREx Class D amplification, supported by a dual mono layout, balanced preamplifier stage, dedicated analog volume controls for each channel, and a power supply designed to keep the amplifier stable under load.

What remains unique is the overall Evo formula: serious power, StreamMagic Gen 4, HDMI eARC, MM phono, balanced XLR input, subwoofer control with crossover and optional high pass filtering, dual speaker zones, headphone output, and a large color display in a chassis that still looks like something designed for a real living space. Cambridge has been very good at making hi-fi feel less like homework, and the Evo 300 pushes that idea further without abandoning the people who still care about system building.

What is missing? Bluetooth support is limited to Bluetooth 5.4 with aptX HD, with no Bluetooth LE Audio, aptX Lossless, or LDAC. There is also no built-in moving coil phono stage, no home theater surround processing, and no claim here that it replaces a true separates system for every listener. At $3,999, it also moves the Evo concept into more serious territory, where buyers will expect StreamMagic stability, loudspeaker control, and long-term software support to be nailed down. The hardware looks ready. The app experience has to keep holding up.

Pricing & Availability

The Cambridge Audio Evo 300 will be available in June 2026 through CambridgeAudio.com and approved retailers for $3,999 USD, while international pricing is set at £3,499, €3,999, and HK$32,800.

Related Reading:

Google has been quietly reaching out to Android developers with an offer to buy access to their code. As reported by 404 Media, the company sent emails to a select group of Google Play developers, inviting them to join what it calls a “confidential content offer pilot.” 

The email frames it as a revenue opportunity, saying developers can “get paid for sharing the code powering your apps, as well as your archived projects.” Google adds that developers retain their intellectual property rights and that the license is non-exclusive.

So what does Google actually want the code for?

According to the report, the email never mentions artificial intelligence, but a link buried in it leads to a page titled “partnerships to improve our AI products.” On that page, Google openly states that it is paying for “non-public content in a range of media formats” to improve its AI models.

Connecting the dots isn’t hard. Google’s Gemini is excellent at image and text generation but has been falling behind in AI coding tools, while Anthropic has ridden the success of Claude Code to a valuation higher than OpenAI. 

OpenAI has also launched its own Codex app, focusing on developers. At the recently concluded Google I/O, the company showcased its Antigravity 2.0 IDE that can create entire apps. 

It seems that Google wants to train its AI with real code to improve its coding capabilities, so it can compete with the likes of Claude Code and ChatGPT’s Codex. Buying real-world code from developers is a shortcut to closing that gap.

Is there anything wrong with this?

While the long-term impact can be detrimental to developers, this approach from Google is not inherently wrong. At least it’s better than training AI on hundreds of thousands of books and online publications without permission, which is something most AI companies have done.

Developers retain their IP, the license is non-exclusive, and they get paid. That said, the lack of transparency in the email is worth noting. Framing an AI data acquisition program as a simple “revenue opportunity” without mentioning AI at all feels like Google is hoping developers won’t ask too many questions.