Google’s Pixel 10 Pro Set to Upgrade PWM Dimming Rate to 480Hz

The imminent release of Google’s flagship Pixel 10 series is generating buzz as leaks about the devices trickle in.

Notably, reports suggest that the Pixel 10 Pro may see an upgrade in its Pulse Width Modulation (PWM) dimming rate to 480Hz, a step up from the current 240Hz seen in earlier models.

This enhancement mirrors similar updates made by Samsung with its Galaxy S24 and S25 series. However, it appears that this upgrade may not extend to the standard Pixel 10 model.

Google’s Pixel 10 Pro Set for Upgraded PWM Dimming Rate

Despite the increase to 480Hz, industry experts remain skeptical about whether this adjustment will satisfactorily address existing issues related to flicker sensitivity.

The current rate still falls short of meeting the IEEE standards, which recommend a minimum PWM rate of 1920Hz to alleviate flicker discomfort. Competing devices from brands like Honor have even pushed the envelope further, achieving PWM rates of up to 4320Hz.

In discussions with Android Authority, it was revealed that Google is reportedly aware of these flicker sensitivity issues and is investigating potential enhancements for the Pixel 10 series.

There are hints suggesting the inclusion of an additional accessibility setting aimed at users with sensitive eyes, which could further improve the PWM dimming rate. Phones from manufacturers like OnePlus and Motorola have already adopted similar features, aiming to ensure a more comfortable viewing experience.

Despite these hopeful signals, concerns remain. Users with flicker sensitivity, such as the author of the report, express ongoing discomfort while using Pixel devices due to their high brightness levels coupled with lower PWM rates.

As anticipation builds for the Pixel 10’s announcement in the coming weeks, it remains to be seen if Google can effectively address these criticisms and improve the overall user experience.

Microsoft says Windows 11 is faster than Windows 10: the numbers used to prove it are highly questionable

Microsoft has just made a bold claim: Windows 11 is up to 2.3 times faster than Windows 10, according to its latest blog post. But when we look closely at the benchmarks used to support that statement, serious doubts arise. The comparison, which seems scientific at first glance, reveals significant flaws in testing conditions, calling into question the credibility of the results.

Benchmarking different hardware, not operating systems

Microsoft based its claims on Geekbench 6 multi-core scores, but the tests did not compare the two systems under the same conditions. Windows 11 was tested on devices with newer processors, including the Intel Core i7-1355U, while Windows 10 was run on much older chips like the Intel Core i7-8750H. That’s a seven-year gap between CPUs—a massive difference that directly skews the outcome, making it impossible to isolate the OS’s actual performance improvements.

Battery life claims follow the same pattern

Microsoft also said that Windows 11 delivers up to 2.7 hours more battery life. Once again, the comparison was between different laptops, with no transparency about battery size, RAM, or usage conditions. Such variables drastically influence power consumption, meaning the boost in autonomy may be due to hardware, not software.

A strategy that lacks technical rigor

This appears to be part of Microsoft’s larger strategy to drive adoption of Windows 11. However, comparing brand-new devices with outdated ones is misleading and doesn’t provide a fair evaluation of the software itself. Users deserve better than marketing tactics disguised as performance data.

Do YouTube’s new AI features make sense?

Google continues to integrate artificial intelligence across its platforms, and YouTube is the latest to receive a significant AI boost. With Premium users in the US now testing new AI-powered features, Google hopes to streamline content discovery. But as helpful as these additions may seem, they raise questions about viewer engagement, content ownership, and the real value of automation on a platform built around creativity.

AI carousels aim to simplify searches

YouTube’s AI-powered carousels offer quick access to relevant clips from various videos when searching for topics like travel, shopping or activities. Instead of browsing multiple full videos, users get a curated selection of highlights and summaries, complete with brief descriptions.

The feature is currently limited to Premium users in the US, but its potential impact is global. By surfacing short video clips without full views, creators may see a dip in watch time and engagement. This could spark backlash similar to what Google’s AI Overviews caused in Search, where websites lost traffic due to users not needing to click through.

Conversational AI brings interactive search to videos

In parallel, YouTube is expanding a chatbot-style assistant that answers questions about video content. Premium users can already ask for more information, recommendations, or even test themselves on academic material. Now, some non-Premium users in the US will gain access too.

The goal is to help viewers better understand or explore the videos they’re watching, but this raises a new dilemma: will users rely on AI summaries instead of diving into full content? The convenience is obvious, but the cost to creators and the platform’s engagement metrics remains unclear.

Is WhatsApp’s unread message summary really private?

Meta’s latest feature for WhatsApp has sparked questions about privacy in the age of AI. The new “unread message summary” lets Meta’s AI create recaps of group chats we haven’t opened, helping us catch up without scrolling through hundreds of messages. But WhatsApp’s long-standing promise of end-to-end encryption raises an obvious concern: how can an AI read our messages without breaking encryption?

A breakthrough in encrypted AI processing

Meta has introduced a technology called “Private Processing” that allows AI to summarize chats without accessing message content in the traditional sense. The process happens inside a sealed virtual machine, a kind of secure hardware chamber where messages are decrypted and summarized — but then instantly deleted.

The messages never leave the encrypted environment and are never stored. If anyone attempts to intercept the process, the system automatically shuts down, preserving privacy. Even Meta itself can’t read the messages, and summaries are not linked to user identity or used for ads.

Balancing usefulness and trust

Still, this innovation arrives at a time when Meta’s track record on privacy remains under scrutiny. From Cambridge Analytica to past data leaks, the company’s promises don’t always inspire confidence. Experts warn that today’s optional summaries could open the door to future features like AI recommendations or ads, even if anonymized.

Setting the tone for private AI

WhatsApp’s move could set a standard for AI use in encrypted environments. Apple has already launched Private Cloud Compute, and other platforms like Signal and Telegram may follow. Meta’s solution shows how advanced AI features can coexist with strong privacy—at least in theory.

iOS 26 call filter ends phone spam: Finally, a real solution

Apple’s iOS 26 introduces a long-awaited solution to unwanted phone calls, giving users a powerful new feature in the Phone app that feels like a personal assistant built right into their iPhone. With spam calls increasing in volume and sophistication, iOS 26’s call filter could be one of the most useful updates we’ve seen in years.

Siri now screens calls for you in real time

With the new iOS 26 call filter, Siri automatically answers unknown numbers on your behalf. Instead of your phone ringing, you’ll get a silent notification. Tap it, and you’ll see a live transcription of what the caller is saying as Siri asks who they are and why they’re calling.

If it’s a robot or spam, the call never even gets through. If it’s someone real, you can read what they say, decide to answer, hang up, or type a reply that Siri will read aloud for you. It’s like having your own virtual receptionist.

How the feature changes daily use

The iPhone becomes smart enough to distinguish between known and unknown numbers, prioritising calls from your contacts or recent message threads. Everything else goes through Siri’s filter. The result: you only deal with the calls that matter.

After each filtered call, a detailed transcript is saved in the Phone app, just like a voicemail. This way, even if you don’t answer, you won’t miss important information. The feature is enabled from Settings > Phone > Call Filter and works without requiring the caller to have any special software.

For many users, this marks the end of the anxiety of picking up unknown calls, wondering if they might be important. With Siri handling the screening, iOS 26 delivers a feature that finally puts control back in our hands.

Most ChatGPT users make this mistake: Three tips to avoid it

Every day, millions of people turn to ChatGPT for help with questions, tasks, and inspiration. But there’s a common mistake that even experienced users continue to make—and it’s quietly holding them back. Here’s how to avoid it and get much better results.

Short prompts are more effective

Long, detailed prompts often confuse the model or dilute the request, leading to vague or off-topic answers. According to a 2024 AI study, prompts with fewer than 12 words perform 22% better. Try replacing lengthy questions like “Can you help me write a blog post about summer road trips with kids?” with “Write tips for summer road trips with kids.” The difference is immediate and noticeable.

Start with a strong verb

Avoid filler phrases like “Can you tell me about…” or “Would you mind explaining…” Instead, begin your prompt with a direct action, such as “Compare,” “Summarize,” “List,” or “Rewrite.” For example, instead of “Could you maybe list some good books for teens?”, just say “List good books for teens.” This removes ambiguity and improves precision.

Add constraints in follow-ups

Trying to pack everything into one giant prompt leads to overload and lower-quality answers. It’s much better to build your request step by step. First ask “List 3 laptops under $1 000”, then refine with “Now filter by models with OLED screens.” This method helps the AI stay focused and accurate.

Less is more when it comes to prompting. By shortening your inputs, starting with a verb, and layering your requests, you’ll get faster, sharper and more relevant responses every time.