Sony’s new True RGB TV lineup  – there are two models, the Sony Bravia 7 II under review here and the step-up Bravia 9 II – is in some ways the most eagerly awaited TV launch of the decade. That’s down to Sony demonstrating promising prototypes of its RGB backlighting technology for several years in a row, and then taking its sweet time to release actual products, even as the competition hurriedly started shipping their own versions of an RGB TV.

Is Sony’s True RGB worth the wait? I’ll get into that more below. But before I do, it will be useful to discuss True RGB and the differences between it and other RGB TV variants (e.g., Samsung Micro RGB, LG Micro RGB evo, TCL RGB Mini-LED and Hisense RGB MiniLED). 

A typical mini-LED TV uses a backlight array with blue or white light modules and relies on a color filter to create the red green and blue color components that make up an image.True RGB TVs like the Sony Bravia 7 II, in contrast, use a mini-LED backlight with clusters of independently controlled red, green and blue light modules. The advantage to this design is that the red, green, and blue lighting components are more efficiently transmitted to individual pixels in the TV’s LCD panel, bringing the benefits of extended P3 and BT.2020 color space coverage, along with enhanced brightness and reduced backlight blooming effects.

Of course, since there is not a one-to-one relationship from backlight to pixel, you still need a color filter, in the form of individual red, green and blue sub-pixels on the LCD panel that create the final colors on each of the 8 million plus pixels on the screen.

While most RGB TVs take a similar approach, Sony’s advanced processing and control algorithms enable the TV to deliver full red, green and blue backlighting even when displaying complex images with a range of colors, resulting in consistent color saturation and a reduction of the “color crosstalk” other RGB TV designs are subject to.

Sony has effectively demonstrated the advantages of True RGB over the RGB LED competition in press demos that eCoustics has attended. LG has also demonstrated the presence of  “color crosstalk” which can occur in certain RGB-backlit TVs (in a comparison designed to showcase the superiority of its OLED models). And TCL acknowledged the same even issue with its own RGB Mini-LED TVs, which it says are inferior to SQD (Super Quantum Dot) Mini-LED models like the TCL X11L and TCL QM8L that use a standard, non-RGB, backlight array.

What Is It?

The Sony BRAVIA 7 II is the company’s entry-level True RGB TV series. It is available in 50-, 55-, 65-, 75-, 85- and 98-inch screen sizes and is priced from $1,599.99 to $8.999.99. The 65-inch model that Sony sent me to review costs $2,599.99, a price that exceeds its entry-level RGB LED competition by a few hundred dollars.

Sony calls the backlight design for its True RGB TVs RGB Backlight Master Drive Pro. Along with this, the Bravia 7 II features Sony’s XR Processor which provides a full suite of XR features such as Clear Image AI (for upscaling high-definition and lower-res content), XR Contrast Booster and XR Motion Clarity. An X-Wide Angle Pro feature of the BRAVIA 7 II’s display panel helps to provide uniform color and contrast when viewing from off-center seats. This alone marks a notable improvement over its predecessor, the BRAVIA 7.

The BRAVIA 7 II supports the Dolby Vision, HDR10 and HLG high dynamic range formats, but not HDR10+. It is also IMAX Enhanced, supporting IMAX Enhanced movies on Disney+ and Sony’s Sony Pictures Core streaming service, including their DTS-X audio tracks sourced from the original IMAX cinematic soundtracks. Calibrated Modes, including Netflix Calibrated Mode, Prime Video Calibrated Mode and Sony Pictures Core Calibrated Mode are included as well, automatically optimizing picture settings when streaming content from these services. My Cinema presets, a new Sony feature, can also be used to optimize picture settings for dark and bright room viewing.

Both the Dolby Atmos and DTS:X immersive audio formats are supported by the BRAVIA 7 II’s built-in speaker array, which pairs dual full-range drivers with tweeters for more accurate directional sound. The set also features 3D Surround Upscaling to create an expansive effect even from two-channel material. The BRAVIA 7 II also has Sony’s upgraded Voice Zoom 3 feature, which uses AI to recognize and either amplify or reduce (an adjustment is provided in the TV’s onscreen menu) dialogue in movie and TV soundtracks.

Sony calls its design for its True RGB TVs “Harmonic Presence.” And while that name might seem somewhat flowery for a TV, the BRAVIA 7 II really does look great in person. A key design element is the TV’s Mirage Stand, which is a steel pedestal stand with a semi-transparent panel that gives it a floating appearance. The rear of the stand also provides effective cable management. Design effort also went into the TV’s remote control, which has textured sides and back and a blue flake pattern embedded in its front surface.

Google TV is used as the BRAVIA 7 II’s smart TV platform, and this version includes the Gemini AI assistant found on several other new TVs. An advantage to Google TV is that, aside from having a wide array of available apps, it can tap your viewing and web search history to make accurate program recommendations. A disadvantage to Google TV is that, like many other smart interfaces, it will show you the occasional ad. Gemini takes content search to the next level by providing a conversational component. (An example: “Hey Google, show me movies directed by David Cronenberg,” followed by, “Show me only the ones from the 1980s.”) And it’s not just for movies – Gemini can be used to research a wide range of topics. Assisting in this task is a built-in mic on Sony’s remote control, and there’s also a built-in mic on the TV for hands-free voice commands.

Sony’s Quick Settings menu gives you easy onscreen access to a range of adjustments including picture mode and brightness. The more in-depth All Settings menu offers up a wide range of picture adjustments such as Auto Local Dimming, Peak Luminance, Advanced Contrast Enhancer, and 2-point and 10-point color temperature adjustments.

Curiously, there are no color management system (CMS) adjustments, a standard feature on many TVs. When I asked Sony why this was the case, I received this reply: “Thanks to our rigorous factory calibration process — backed by our engineering design that accurately calculates and optimizes performance at the production stage — color accuracy and gamma are already dialed in. A simple white balance/greyscale adjustment is all that’s needed to achieve reference picture quality, and everything else falls into line automatically.”

I had expected that the Sony True RGB lineup would continue the trend started by the BRAVIA 3 II earlier this year with its four gaming-friendly HDMI 2.1 ports, but that didn’t turn out to be the case. The BRAVIA 7 II has two HDMI 2.1 ports with 4K/120Hz support and two HDMI 2.0b ports (one with eARC). Other gaming features include Perfect for PlayStation5, which enables ALLM, Auto HDR Tone Mapping and Auto Genre Picture Mode when a PS5 console is connected, and a Game Menu 2 onscreen pop-up menu for making gaming-related adjustments.

Aside from its four HDMI ports, the BRAVIA 7 II provides USB Type-A (x2), Ethernet and RS-232C data connections, an IR receiver input and a combined optical digital audio output/analog center speaker input, with the latter meant for Sony wireless speaker systems that use the TV’s built-in audio as a center channel speaker. There’s also an RF input for the set’s ATSC 3.0 (Next Gen TV) digital TV tuner. When an external antenna is connected to this, both ATSC 3.0 and standard digital broadcast channels can be browsed separately in the Google TV Live TV grid guide.

Go Pro?

The Sony BRAVIA 7 II (and the BRAVIA 9 II) in 65-inch and above screen sizes will also be available in a “Pro” package that includes a longer 3-year full replacement warranty, an upgraded rechargeable and backlit remote control, additional Sony Pictures Core movie credits and an enhanced version of Voice Zoom 3. Hardware for the standard and Pro versions is otherwise identical. Sony didn’t share pricing for the Pro package, which is aimed at the CI (custom installation), channel but it seems to be a nominal increase over the standard offering.

Setup & Viewing Impressions

Before doing any subjective viewing, I  ran a basic set of measurements on the BRAVIA 7 II using Portrait Display’s Calman Color Calibration software. The TV’s ambient light sensor was disabled before I began testing; otherwise, measurements were made in the TV’s Cinema and Standard picture modes with the default settings left intact. Sony’s Professional mode is said to be the most accurate picture preset option, but I found it to be way too dim for SDR viewing at its default settings and I expect most viewers would have the same take. But since it is the most accurate mode, we did include some out of the box measurements in Professional mode for HDR content.

Peak HDR brightness measured on a white 10% window pattern in Cinema Mode was 1,800 nits and 701 nits on a 100% (fullscreen) white pattern. For Professional mode, the results were 1,971 nits and 790 nits, respectively, on the same tests. In Standard mode, peak HDR brightness was 1,554 nits on a 10% pattern and 626 nits for fullscreen.

In SDR (standard dynamic range) tests, the BRAVIA 7 II measured 486 nits on a 10% pattern in Cinema mode and 421 nits on a fullscreen one. In Standard mode, the results were 576 nits for 10% and 372 nits for fullscreen.

Those brightness results were essentially what I expected from the Bravia 7 II, which Sony reportedly stated would be half as bright as the step-up Bravia 9 II (at 4,000 nits). For context, the Sony’s brightness measurements put it in roughly the same range as the TCL QM8L SQD Mini-LED TV, which measured 1,922 nits (10% window) and 708 nits (fullscreen) in Filmmaker Mode when I tested it.

The BRAVIA 7 II’s color gamut coverage in Cinema Mode measured 88.5% for BT.2020 and 98.5% for P3. That BT.2020 result trails the only other RGB TV I’ve measured, the Samsung R95H Micro LED, which yielded 92% coverage (in Filmmaker Mode). For comparison’s sake, the BRAVIA 7 II’s color gamut coverage is closer to the TCL QM8L, which measured 89.3% for BT.2020 and 97.7% for P3 (both in Filmmaker Mode) when I tested it.

In other BRAVIA 7 II measurements, the Delta-E (the margin of error between the test pattern source and what’s displayed on-screen) averaged 1.6 for grayscale and 3.5 for color. The Sony’s color result is slightly higher than the 3.0 Delta-E that’s considered to be the threshold for what’s indistinguishable from perfect to the human eye, and it was also heavily weighted by a 6.4 red error.

Starting off my viewing tests with the Spears & Munsil Ultra HD Benchmark demonstration reel (2,000 nits version), the BRAVIA 7 II displayed excellent color, contrast and detail throughout. High-contrast shots of brightly-lit objects and animals against a pitch black background showed no signs of backlight blooming, and overhead shots of a city at night revealed excellent micro-contrast. Switching to the 10,000 nits version, the Sony’s HDR tone mapping at the default Gradation Preferred setting was excellent, revealing much the same level of highlight detail as in the 2,000 nits version of the demonstration reel.

The disc’s starfield test loops showed the TV’s local dimming to be a bit too aggressive at the Cinema preset’s default Medium setting, obscuring significantly more stars than at the Low setting, but I ultimately found Medium to be my preferred preset when watching movies and TV. The Spears & Munsil disc’s Dots test pattern, which is used to evaluate off-axis picture uniformity, revealed very good color saturation and contrast when viewing from off-center seats, showing the strength of Sony’s X-Wide Angle Pro panel. A bit of blooming was visible when viewing off-axis, but that was mostly an issue when looking at test patterns.

Sony TVs are usually very good at upscaling lower-resolution video, and when I watched a DVD of David Lynch’s Wild at Heart, I was impressed at how clear and detailed the picture looked. (I’ve certainly seen this disc look much worse on other 4K TVs.) As far as I know, the Wild at Heart Blu-ray is out of print, and there’s been no 4K Blu-ray version yet, so the DVD will have to do for now!

Watching a 4K Blu-ray of the James Bond film No Time to Die, a scene that I use as a spot-check for motion handling where 007 walks across a craggy hillside cemetery looked solid and mostly judder-free. I say “mostly” because there was a very slight amount of judder, but it was easily eliminated by adjusting the Smoothness (Film) setting to 1 in the TV’s MotionFlow menu.

I also use the car chase from No Time to Die, a sequence that follows after an explosion at the cemetery, as a spot-check for TV sound, and the Bravia 7 II’s built-in speakers proved to be very much up to the task. Bass effects were solid, and there was a good sense of spaciousness to the movie’s Dolby Atmos soundtrack. Dialogue in this and other movies was also very clear, even without having to engage the TV’s Voice Zoom 3 feature.

To evaluate shadow depth/detail and color, I watched scenes from Spider-Man: Into the Spider-Verse and Alien: Romulus (both in 4K/Dolby Vision) on my Apple TV 4K. In Spider-Man, contrast was strong and there was plenty of shadow detail visible in the dark subway tunnel scenes. Colors in the fight sequences with the Green Goblin also looked bright and bold. I had similar impressions when watching Alien: Romulus: Dark chambers in the abandoned space station were rendered with deep, uniform blacks, and the colored lights of control panels popped with a good degree of intensity.

The Bottom Line

I thought I had tested the ultimate Sony TV when I had a chance to review the Sony Bravia 9 mini-LED, but the new mid-range Sony Bravia 7 II True RGB TV not only matches, but in key ways exceeds that model, specifically its P3 and BT.2020 color gamut coverage.

As you might expect given the True RGB backlight, color is the star of the show here, with the Bravia 7 II consistently serving up clean, rich-looking color. While its peak brightness doesn’t hit the same lofty heights as some other TVs I’ve recently tested, the Bravia 7 II is plenty bright enough even for viewing in a well-lit room, and its low screen reflectivity means you won’t have to worry that much about screen reflections. (With the step-up Bravia 9 II, which costs $1,000 more and features an Immersive Black Screen Pro glare-free screen, you won’t have to worry about reflections at all.)

Is the Bravia 7 II the best TV you can get for the money? If I were a serious gamer, I’d find its two HDMI 2.1 port count limit off-putting, especially when many new-for-2026 TVs feature three or more HDMI 2.1 connections. Also, there’s no support for HDR10+ or the forthcoming Dolby Vision 2 format, which offers numerous important and forward-looking features. Given that the Bravia 7 II is pricier than competing models such as the TCL QM8L and Samsung R85H, I might give those points some consideration.

But the real bottom line here is that everything I watched looked great on the BRAVIA 7 II, my old Wild at Heart DVD included. And with its elegant Harmonic Presence design and Mirage Stand, Sony’s entry-level True RGB TV looks good even when turned off.

Pros:

• Refined local dimming
• Rich, mostly accurate color
• Very good off-axis uniformity
• Impressive brightness
• Clean, detailed upscaling
• Solid built-in audio and sound features
• Elegant design

Cons:

• No HDR10+ or Dolby Vision 2 upgrade support
• Only two HDMI 2.1 ports
• Pricier than mid-range mini-LED TV competition
• Some screen reflections

Our Ratings:

★★★★★★★★★★ Picture Quality

★★★★★★★★★★ Design

★★★★★★★★★★ Usability

★★★★★★★★★★ Sound Quality

★★★★★★★★★★ Features

★★★★★★★★★★ Value

Where to buy

BRAVIA 7 II

BRAVIA 7 II PRO

Also available from: B&H, Sony USA, Value Electronics

Related Reading

GoPro warned on Monday that there is “substantial doubt about the company’s ability to continue as a going concern.” The action-camera maker reported a 26% revenue decline in Q1 and expects to breach several loan covenants. Shares fell as much as 14%.

The cause is memory. GoPro said its earnings forecast has been “significantly impacted” by an 80% to 115% increase in memory prices. In April, suppliers informed the company of a planned reduction in memory supply that would further reduce forecasted sales. The same DRAM reallocation that is killing the cheap smartphone is now threatening to kill GoPro.

The mechanism is the one we detailed last week. Samsung, SK Hynix, and Micron have redirected wafer capacity from consumer DRAM to high-bandwidth memory for AI data centres. HBM margins run at 70% or higher. Consumer DRAM margins sit between 20% and 30%. The memory makers chose the higher-margin customer. Everyone else pays more or gets less.

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!

GoPro does not have the purchasing power to absorb the price increase. It is not Apple, which can negotiate quarterly contracts and pass costs onto consumers buying $1,000 phones. It is a sub-$1 billion revenue company whose products sell for $300 to $500 and depend on commodity memory to store high-resolution video. When memory costs double, the product becomes unprofitable.

The company has received waivers from its lender after failing to comply with loan covenants. It does not expect to have enough liquidity to meet obligations if default provisions are triggered and outstanding debt becomes due. It has a $50 million second-lien facility from Farallon Capital Management and a revolving credit facility with Wells Fargo as agent.

GoPro has engaged advisors to evaluate strategic alternatives including a potential sale or merger. It is also exploring opportunities in defence and aerospace for “new markets and product categories.” The company already announced plans to cut 23% of its global staff in April.

The defence pivot echoes Faraday Future’s robotics pivot: a consumer electronics company under financial pressure reaching for a higher-margin, government-funded market where the competitive dynamics are different. Whether GoPro’s ruggedised camera expertise translates into defence contracts is unproven.

The only near-term supply relief is coming from China. ChangXin Memory Technologies’ DRAM has been spotted inside Corsair’s retail DDR5 kits. But CXMT is also planning to convert 20% of its capacity to HBM because the margins are irresistible. The consumer memory shortage is structural, not cyclical.

The memory crisis is visible across consumer electronics. The Asus ROG NUC 16 costs $1,200 more than last year’s model, partly due to DDR5 prices. Dell hiked laptop prices 15-20% in December. Apple agreed to pay Samsung a 100% premium on LPDDR5X for the iPhone. These companies can absorb the cost. GoPro cannot.

GoPro was founded in 2002 by Nicholas Woodman. It went public in 2014 at a $3 billion valuation. The company popularised the action camera category and built a brand that became synonymous with extreme sports and adventure content. Its share price peaked above $90 in 2014. It trades below $1 today.

The going-concern warning makes GoPro the most visible corporate casualty of the AI memory reallocation. It will not be the last. Any consumer electronics company with thin margins, limited purchasing power, and dependence on commodity DRAM is facing the same calculus. The AI boom created enormous wealth for three memory makers and the hyperscalers they supply. GoPro is on the other side of that equation.

On March 31, I received an email from Norse Atlantic Airways. The $940 flights for my upcoming round trip to Rome had been canceled, it said, and I had 14 days to request a refund.

At first, I didn’t panic. That began to change when the company’s refund request page wouldn’t load on two browsers across three devices. After Norse didn’t respond to several emails, I looked for a phone number. There wasn’t one. On Reddit, I found dozens of posts about Norse’s allegedly haphazard customer service.

The same day, I filed a public records request with the Federal Trade Commission, which I hoped would give me a better idea of how common this experience was. I eventually received around 75 detailed complaints from people who had bought or tried to buy tickets from the airline. Many described a customer service operation in which the inability to get in touch with a human created a vacuum that scammers appeared happy to step into. Of the 41 complaints that reported a dollar figure, 21 claimed they lost more than $1,000.

Norse Atlantic Airways does have human customer service workers, but in recent years, the airline has leaned into a tech-forward approach, deploying AI agents to help power its operation.

“Technology will help us have a higher level of availability and customer support, while still maintaining low fares for more people to enjoy travel between continents,” Bård Nordhagen, the company’s chief customer and communications officer, tells WIRED.

Yet if what I and dozens of other people experienced is any indication, this version of customer service is time-consuming, frustrating, and at times expensive.

The Future Is Now

Norse Atlantic Airways, which was formed in February 2021, has described itself as a “modern, long-haul, low-cost airline” with a “lean” workforce. Early on, it implemented a tool from the customer service technology company Sprinklr that created a “unified” inbox of customer service queries. (Based on archives of the company’s website, it doesn’t appear to have ever listed a customer service number.)

In January 2025, the AI company Kindly wrote a blog post detailing how it developed a chatbot for Norse alternatingly called “Odin” or “Odin’s Wingman.” Norse also removed the customer support email from its support page in order to make Odin the “primary support channel,” according to the Kindly blog post.

By January 2026, Norse had “sunset” the chatbot and replaced it with its current AI agent, Freya. Delight.ai, the company that developed Freya, said that the airline’s no-human-intervention inquiry resolution rate “rose from 60 percent to 80 percent” within two weeks of its introduction.

“We see the future of our customer support team as AI agent managers,” Norse’s chief product officer, Alf Lim, said in a Delight.ai blog post. Lim added that Freya is a “core part of the team” at Norse.

According to the blog, Freya would allow Norse to “upskill” its customer support unit into these AI agent managers, which are described as “specialists who continuously optimize, train and step in when human-touch is required.”

Nordhagen tells WIRED that Freya has been a success and now manages “99 percent of inquiries from passengers.”

A Scammer’s Paradise

Many of the FTC complaints shared a common theme: A person, needing to change their flight or adjust their booking, searched online for the Norse Atlantic Airways phone number. Eighteen of the FTC complaints explicitly claimed that the person was scammed after they Googled Norse’s customer service information and found scam websites and phone numbers in the results.

In some cases, customers claimed they were told they owed money for a flight they thought they already paid for. Other times, they said they were told that they had to pay an exorbitant fee in order to make a change to their itinerary.

Sony is finally setting release dates for two PlayStation accessories that have been in the “coming soon” category since their debut last year. The company has confirmed that preorders for its FlexStrike wireless fight stick and its unusual 27-inch gaming monitor with a built-in DualSense charging hook will begin this June, with both products arriving later this summer.

For fighting game fans and desk-bound PlayStation players, these are arguably two of Sony’s most interesting hardware experiments outside the PS5 itself.

FlexStrike arrives just in time for fighting game season

Sony’s first wireless fight stick, the FlexStrike, will officially launch on August 6. The accessory is set to arrive alongside the growing excitement around Marvel Tōkon: Fighting Souls, giving competitive players a new premium controller option for PS5 and PC.

Unlike traditional arcade sticks that can be cumbersome to carry around, Sony is pitching FlexStrike as a tournament-friendly device. It comes with a built-in rechargeable battery and even includes a dedicated sling carry case, making it easier to haul between local meetups, esports events, or simply a friend’s house. Preorders open on June 12, with a $200 price tag that places it squarely in enthusiast territory. That’s not cheap, but fight stick players have historically shown a willingness to invest in specialized gear if it offers reliability and convenience.

Sony thinks your gaming setup has moved beyond the living room

The second accessory may be even more intriguing. Sony’s new 27-inch gaming monitor is clearly aimed at players who have shifted from couch gaming to dedicated desk setups. The display features a 1440p IPS panel with variable refresh rate support and can hit 120Hz when paired with a PS5 or PS5 Pro. Connect it to a capable PC or Mac, however, and the refresh rate jumps to 240Hz.

What makes the monitor stand out isn’t necessarily its specs, though. It’s the small built-in charging hook designed specifically for a DualSense controller. Sony will open preorders for the monitor on June 5 through PlayStation Direct and Best Buy in the U.S., with a retail price of $350. The display is scheduled to launch on August 27 in the U.S. and Japan.

Both products signal Sony’s growing interest in gaming beyond the traditional console-and-TV setup. Whether you’re grinding ranked matches in a fighting game or building out a dedicated PlayStation desk station, Sony clearly wants a piece of that ecosystem too.

Children born after 2013 are the first generation to grow up fully immersed in digital systems, which weren’t designed with them in mind. One‑third of the world’s Internet users are younger than 18, according to UNICEF, yet these systems shaping their daily lives were built for adults. They were optimized for engagement and designed long before people understood how profoundly digital environments influence children.

For engineers and technical professionals, online safety is not an abstract policy debate. It is a design challenge that demands rigor, systems thinking, and ethical foresight.

Governments around the world are also beginning to recognize the problem. Policymakers from across Australia, Brazil, the European Union, Indonesia, and the United States are responding to risks engineers have long understood: Addictive features, inappropriate content, opaque data practices, and algorithmic systems shape user behavior in ways that their creators did not fully predict. For years, technology moved faster than governance. Now governance is trying to catch up.

Global Shift Toward Design Reform

The European Union and the United Kingdom have been among the first to act, embedding age‑appropriate digital design into their broader children’s rights agenda. Drawing on IEEE expertise and global best practices, Indonesia is the first country in Asia, and Brazil is the first country in Latin America, to adopt age-appropriate design regulation. Australia is aiming to limit access to harmful content and addictive design features through age restrictions on certain platforms. And in the United States, in addition to federal efforts, states including California, New York, and Utah are enacting approaches including age-appropriate design principles.

Across these efforts, a shared realization is emerging. Protecting children online is not simply about filtering content or adding parental controls. It requires rethinking the architecture of digital systems regarding how data is collected, how algorithms make decisions, how interfaces influence attention, and how AI interacts with the developing minds of young users.

Engineers and technical professionals understand that design choices are never neutral. They encode values, incentives, and assumptions. When the user is a child, those choices carry greater weight.

This is where IEEE’s work becomes more essential.

Protecting Children Online

For more than a decade, IEEE has been building technical and ethical foundations for safer digital experiences. The first IEEE standard on age-appropriate design in 2021 marked a turning point. It offers a structured, principled approach to designing with children’s rights in mind. The Institute’s 2022 article “Use a New IEEE Standard to Design a Safer Digital World for Kids” highlights how the standard helps translate those principles into engineering practice.

Today the IEEE Standards Association’s (SA) Trustworthy Digital Experiences portfolio provides a practical, technically grounded framework for governments and industry. Spanning ethical design, data governance, algorithmic transparency, and child‑focused digital well‑being, it has already initiated discussions with government stakeholders around the world. This work helps bridge the gap between engineering realities and policy ambitions.

No single country can solve these challenges alone. Many policymakers lack access to the combined expertise in technology, governance, and children’s rights needed to act quickly and effectively. This collaborative effort helps close that gap.

The stakes are high. Without coordinated action, public policy will continue to lag behind technology, leaving children exposed to risks that could have been mitigated through thoughtful design. But with the right frameworks, governments can ensure digital systems respect children’s rights, support healthy development, and promote well‑being.

IEEE’s emerging standards and collaborative technology policy work offer a path forward. By grounding national efforts in evidence‑based, rights-aligned design principles, IEEE is helping governments move from reactive regulation to proactive, coherent, and globally informed strategies for protecting children online.

Safeguarding childhood in the digital age is both a moral imperative and an engineering challenge. And IEEE is helping to lead the way.

—Mary Ellen Randall

IEEE president and CEO

Please share your thoughts with me: president@ieee.org.

This article appears in the June 2026 print issue.

Across the frontier labs, the highest prompt injection figures published this spring are Anthropic’s. Point a red-teamer at its newest model in a browser, and the attacker hijacked it 31.5% of the time before safeguards engaged. OpenAI, Google, and Meta never gave security leaders a comparable number to set beside it. That figure looks like a liability. In this comparison, it is the opposite. It’s the one solid piece of ground.

Four frontier labs each shipped a prompt injection disclosure, and no two match. Anthropic put 244 pages and four agentic surfaces on the table on May 28. OpenAI reported one surface, connectors. Google moved the subject out of the model card and into a separate safety framework. Meta shipped no closed-model card at all. The Cross-Vendor Prompt Injection Disclosure Grid below maps what each lab tested, what each one measured, and the four places a side-by-side comparison falls apart.

A prompt injection hides a malicious instruction in something an agent reads, a web page, a document, or a tool result. One planted line can exfiltrate records or fire off actions nobody approved, and these cards are a buyer’s only first-party evidence.

There is no industry standard for measuring any of this, and that is the root of the problem. Carter Rees, VP of AI at Reputation, told VentureBeat that prompt injection breaks the assumption that every legacy tool was built on. “A phrase as innocuous as, ‘ignore previous instructions’ can carry a payload as devastating as a buffer overflow, yet it shares no commonality with known malware signatures.” With no shared signature to scan for, each lab built its own yardstick, and the results do not line up.

Adam Meyers, Senior Vice President of Counter Adversary Operations at CrowdStrike, said that the exposure is now the buyer’s to manage. “As you implement AI, it increases your attack surface, so now you have to be able to protect those AI models against adversary misuse or data poisoning or prompt injection.” CrowdStrike’s own frontline data shows the threat side is not standing still. In its 2026 Financial Services Threat Landscape Report, released in May, the company reported adversaries using AI to compress the time from initial access to impact faster than legacy defenses can respond.

Anthropic measured four surfaces. The numbers swing by an order of magnitude depending on which one you read.

The Opus 4.8 card does what others do not: It breaks prompt injection out by surface, and the spread is the story.

Put the model in a coding environment, and an adaptive attacker from Gray Swan’s Shade tool got through on 7.03% of single attempts with thinking on. Safeguards pulled that to 2.09%.

Move the same class of attack into a browser, the surface behind Claude in Chrome and Claude Cowork, and the floor gives way. Anthropic put professional red-teamers on 129 web environments held out from training and printed every result in Table 5.2.2.4.A on page 81 of the system card. Per-attempt is the share of all injection attempts that got through across 129 environments at 10 tries each. Per-scenario is the harder cut, the share of environments where at least one try landed.

Read down the per-attempt column without safeguards, thinking on, and the raw rate drops with each generation, from Sonnet 4.6 at 50.7% to Opus 4.8 at 31.5%. The lowest in the table, 5.9%, belongs to Mythos Preview, which nobody can buy yet. Turn safeguards on, and Opus 4.8 drops to 0.5%. Turn thinking off and it drops to zero across all 129 environments.

OpenAI measured one surface, with attacks it already knew.

The GPT-5.5 card, published April 23 and updated April 24, handles prompt injection in one place, a single section on robustness to known attacks against connectors. OpenAI reports it as a robustness score where higher is better, the inverse of an attack success rate. GPT-5.5 came in at 0.963, down from 0.998 for GPT-5.4-thinking. That one figure is the whole disclosure.

Anthropic tested four surfaces against an adaptive attacker that rewrites its approach based on what the model does, then ran a one-week bug bounty where red-teamers tried to break the model live. When the coding results came back worse than Opus 4.7, the card said so.

Lay the 0.963 next to the 31.5%, and they look like they belong on a scoreboard. They do not. One is a robustness score against known attacks on one surface. The other is a per-attempt attack success rate across 129 browser environments against an attacker that adapted in real time.

Google and Meta never put the number in the card at all

Google’s Gemini 3 files prompt injection under mitigations, and the launch materials describe stronger resistance with no number attached. The Frontier Safety Framework report does run red teaming, but across its capability domains, and prompt injection is not one of them. No model card, no framework page, no per-surface number a buyer can lift into a risk review.

Meta ships open weights with no closed-model card. Prompt injection defense sits in a separate stack, Purple Llama’s LlamaFirewall. A PromptGuard 2 classifier and an AlignmentCheck auditor, run against the public AgentDojo benchmark and its 97 tasks, cut attack success from 17.6% with no defense to 1.75% combined. Real numbers. They grade the guardrails on a public benchmark, not the model on a deployment surface a security team would recognize.

The Cross-Vendor Prompt Injection Disclosure Grid

The grid below works on any frontier model security teams are weighing. Each row marks a place where the four labs are split. Each split is where a quick comparison breaks. The Anthropic figures come from the Opus 4.8 system card. Everything for the other three comes from each vendor’s published safety documentation.

Five factors security teams need to consider now

Anthropic tested four surfaces and printed every number. OpenAI tested one. Google printed no per-surface rate. Meta graded its guardrails, not the model. The four disclosures do not add up to a comparison. These five steps build one.

Pull every agent you have deployed or scoped and tag each by the surface it touches, browser, code, connectors, or desktop. Anthropic’s rate for Opus 4.8 runs 2.09% on coding and 0.5% on browser. A blended number covers neither. Pull the vendor’s published rate for your specific surface. If the vendor never published one, treat it as untested.

Send the Cross-Vendor grid to every vendor under evaluation. A 0.963 connectors score and a 31.5% browser rate were never on one scale. Demand a per-surface attack success rate, raw and safeguarded, with the attacker methodology named. The blank cells are the surfaces with no first-party evidence.

Confirm in writing which number your integration gets. Anthropic’s 0.5% comes from Claude in Chrome and Cowork with the full safeguard stack. On the API, the model ships without them. Do not accept a product number for an API deployment.

Add two clauses to the RFP. The vendor tested with an adaptive attacker that rewrites payloads against the model, and someone outside the company tried to break it. Anthropic ran Gray Swan’s adaptive Shade tool and a one-week paid bounty. OpenAI tested known attacks on one surface. Adversaries do not submit known payloads.

Run your own injection test before any agent ships. Vendor numbers come from vendor environments with vendor system prompts. Your stack has its own prompts, permissions, and data access. Set a pass threshold. Anything above it does not go live.

The bottom line. No standard exists for this yet. A vendor’s number tells you what it chose to measure. Your own red team tells you what you are exposed to.

Systems

China’s support is greater relative to semiconductor industry revenue

A report from the Organization for Economic Co-operation and Development (OECD) has found that semiconductor firms based in the United States received more government support than those based in any other region.

However, support for China’s chip industry was larger relative to the revenue generated by Chinese semiconductor firms, reaching close to 10 percent of sales in the early 2020s.

The OECD – a forum for members espousing the market economy and democracy – said the global semiconductor market was worth $631 billion in 2024. It expected continued growth on the back of investment in datacenters, artificial intelligence, and autonomous driving. Its measure of the market includes chip design, manufacturing, testing and packaging, but not manufacturing equipment such as photolithography machines.

Firms based in the United States and Asia (eg Japan, Korea, and Taiwan) have long been the key players in the semiconductor sector, with Asia’s role growing in importance as part of the supply chain was relocated there. Asia has, over the last two decades, become a global center for chip manufacturing and trade, although the United States maintains an important role in high-value segments of the supply chain, including in chip design. The sample of firms covered by the OECD MAGIC database thus includes a relatively large number of firms based in Asia and the United States, as well as large actors based in Europe, which largely serve the automotive industry. The sample is estimated to cover between 64 percent and 83 percent of global sales, depending on the year and how the sector’s scope is defined.

“In absolute terms, firms based in the United States were the largest beneficiaries of government support, which notably includes the support these firms received in other jurisdictions in which they operate (eg in Asia), as well as the introduction of new subsidy programs in the United States. Subsidies to firms based in the OECD Asia-Pacific region also expanded steadily throughout the period for similar reasons. While subsidies to semiconductor manufacturers based in China have been modest in absolute terms, they represented a significant amount relative to their sales, reaching close to 10 percent of revenue in the early 2020s,” the report said.

The OECD argued that China’s relatively large support reflected its long-standing support for its semiconductor industry, including the 2014 Guideline for the Promotion of the Development of the National Integrated Circuit Industry. It also reflected growing restrictions imposed on exports of semiconductor technology by trading partners beginning in 2018, the report said.

The report measures tax concessions, grants and subsidized borrowing as methods governments use to support semiconductor firms. It does not include government equity.

In August last year, the Trump administration took a 9.9 percent equity stake in struggling chipmaker Intel, using $5.7 billion in previously awarded but unpaid CHIPS Act grants as part of an $8.9 billion investment agreement.

The administration has also tried to bolster domestic chip manufacturing with its tariff regime, although, given the time it takes to build a fab plant, it might take years for the policy to pay off. ®

Tencent has announced that PayPal users will be able to make cashless payments in China by scanning QR codes through WeChat Pay’s merchant network. The integration, which will be available to US-based PayPal users first with more markets to follow, connects PayPal’s 400 million-plus user base to the payment infrastructure that covers virtually every taxi, restaurant, and shop in mainland China.

The move addresses a specific pain point. China’s economy runs on mobile payments, with WeChat Pay and Ant Group’s Alipay processing trillions of dollars in transactions annually. Foreign visitors who arrive without access to either platform often find themselves unable to pay for basic goods and services, since many Chinese merchants no longer accept cash and few accept international credit cards at the point of sale. Both WeChat Pay and Alipay have allowed foreigners to link international bank cards since 2019, but adoption has been limited by a cumbersome onboarding process that requires downloading Chinese apps and navigating interfaces designed for domestic users.

Why PayPal matters

PayPal integration sidesteps the onboarding problem entirely. A US tourist arriving in Beijing does not need to download WeChat, create an account, or link a bank card to a Chinese payment platform. They scan a WeChat Pay QR code with their existing PayPal app, the transaction processes through WeChat Pay’s merchant network, and the charge appears on their PayPal balance or linked card. The friction reduction is significant for a demographic that currently struggles with China’s cashless infrastructure.

Tencent is also waiving transaction fees for first-time users who link international bank cards directly to WeChat, a separate incentive aimed at encouraging deeper integration beyond the PayPal pathway. The company reported that foreign traveller transactions in China jumped nearly 80% year on year in the January-to-April period of 2026, suggesting that China’s broader tourism push is already driving payment volume growth.

China’s tourism strategy

The PayPal deal is part of a coordinated effort by Beijing to attract more foreign tourists after the pandemic-era collapse in international arrivals. China has expanded visa-free access to travellers from dozens of countries, including the UK, Spain, and Australia, though US travellers still require a visa except for brief transits to third countries.

The strategy is working by the numbers. Foreign visitors to China, excluding those from Hong Kong and Taiwan, surpassed 35 million in 2025, exceeding the pre-pandemic record of nearly 32 million set in 2019. Tourism contributed more than 4% of China’s GDP in 2024. China’s approach to technology adoption has created an economy where digital infrastructure is more advanced than in most Western countries, but that very advancement creates barriers for visitors who are not part of the domestic digital ecosystem.

Gary Ng, a senior economist for Asia Pacific at French bank Natixis, said the move aligns with a global trend of payment platform integration through mutually recognised cross-border QR codes. The same dynamic is playing out in Southeast Asia, where Singapore, Thailand, and Malaysia have established bilateral QR code payment links.

Limited near-term impact

The practical significance of the PayPal integration depends on the volume of US travellers to China, which remains relatively low compared to pre-pandemic levels and is further constrained by the ongoing visa requirement. Ivan Su, a senior equity analyst at Morningstar, said the initial impact may be limited in terms of overall benefit for Tencent given those volumes.

The competitive dynamic between WeChat Pay and Alipay adds another dimension. As fintech platforms globally expand into cross-border payments, both Chinese platforms are racing to be the preferred gateway for foreign visitors. Alipay has its own international partnerships and has been more aggressive in marketing to tourists through in-app translation features and curated city guides. Alibaba’s broader technology ambitions give Alipay a parent company with deep resources to invest in cross-border payment infrastructure.

For Tencent, the PayPal deal is less about immediate revenue and more about positioning WeChat Pay as the default payment rail for international visitors. Chinese platform companies are navigating a regulatory environment that encourages international engagement while maintaining domestic control over data and financial flows. Connecting PayPal to WeChat Pay’s merchant network threads that needle, giving foreign users access to the payment infrastructure without giving them direct access to the underlying Chinese financial system.