gHacks Technology News

OpenAI is recruiting a new Head of Preparedness, a senior role focused on mitigating risks tied to advanced AI systems. The position comes with a reported annual compensation of roughly $555,000, plus equity, underscoring how seriously the company says it is taking emerging safety concerns.

OpenAI CEO Sam Altman publicly acknowledged the opening in a social media post, describing the job as demanding and warning that it would be "stressful." According to the listing, the role is responsible for reducing harms linked to AI capabilities, including risks to mental health, cybersecurity, and biological misuse.

The hiring push comes as AI systems become more capable and harder to predict. Altman said models are now powerful enough to create "real challenges," especially when it comes to abuse scenarios where attackers could repurpose AI tools faster than defenses can adapt.

Preparedness work at OpenAI is not new, but the role has shifted. The company's previous head of preparedness, Aleksander Madry, was reassigned last year to a position focused on AI reasoning, with safety becoming a secondary responsibility. The new opening restores preparedness as a standalone executive function.

The timing is notable. A growing number of companies are flagging AI-related reputational and operational risks in regulatory filings, and OpenAI itself has acknowledged that some upcoming models could pose elevated cybersecurity threats. The company has said it is expanding monitoring systems and training models to refuse requests that could compromise security.

OpenAI has also faced public scrutiny over mental health concerns. The company has been named in multiple lawsuits alleging harmful interactions with ChatGPT, and external investigations have documented cases where users experienced severe distress during extended conversations with the system. OpenAI has responded by updating how ChatGPT handles sensitive topics, adding crisis resources, and funding research into AI and mental health.

For OpenAI, the Head of Preparedness role is meant to sit at the intersection of technical capability and real-world impact. The position is tasked with anticipating how new systems might be misused and shaping release strategies that limit harm without halting development entirely.

The role does not signal a slowdown in OpenAI's ambitions. Instead, it reflects an acknowledgment that as AI systems approach higher levels of autonomy and reasoning, safety planning must move earlier in the development process and carry more authority.

Thank you for being a Ghacks reader. The post OpenAI Is Hiring a Head of Preparedness to Address AI Safety Risks appeared first on gHacks Technology News.

Researchers from University of Pennsylvania and University of Michigan have built what they describe as the world's smallest fully autonomous robots. The devices are small enough to operate at the scale of biological cells, measuring roughly 200 × 300 × 50 micrometers, about one-tenth the width of a millimeter.

Despite their size, the robots are self-contained. Each unit can move, sense its environment, perform basic computation, and respond without external control, tethers, or magnetic guidance. According to the researchers, the robots are also inexpensive to manufacture, costing around one cent per unit.

The work was led in part by Marc Miskin, an assistant professor of electrical and systems engineering, who noted that these robots are orders of magnitude smaller than previous micro-robotic systems. Their size places them in the same range as many microorganisms, which opens up research possibilities that larger robots cannot reach.

The findings were published in Science Robotics and Proceedings of the National Academy of Sciences, detailing how the robots integrate power, logic, and movement into a structure smaller than a grain of salt. The robots are designed to survive and operate autonomously for extended periods, potentially lasting months under the right conditions.

Researchers suggest long-term applications could include tracking the behavior of individual cells, studying microscopic environments, or assisting in the construction of microscale machines. Because the robots function at the same physical scale as biological systems, they may eventually be able to navigate tissue or lab-grown environments that are inaccessible to existing tools.

This research remains experimental. There is no consumer-facing application, and practical use outside laboratory settings is still theoretical. The work represents a technical milestone rather than a deployable technology, highlighting how far miniaturization and autonomous control have progressed rather than signaling an imminent product or service.

Thank you for being a Ghacks reader. The post Researchers Build Autonomous Robots Smaller Than a Grain of Salt appeared first on gHacks Technology News.

Google ended official OS updates for the Pixel 4 and Pixel 4 XL in 2022, but the devices are not stuck there. The LineageOS team has added official support for the Google Pixel 4 and Pixel 4 XL in LineageOS 23, bringing Android 16 to Google's 2019 phones.

This is a step up from earlier unofficial builds. Official status means the maintainers consider the ROM stable enough for daily use, with core features working reliably. For users still carrying a Pixel 4 as a main or backup phone, this extends the device's usable life well beyond Google's support window.

LineageOS 23 went official earlier this year and is based on Android 16. Alongside the platform upgrade, the release includes refreshed system apps. The Aperture camera app has been rewritten for better performance and gains support for features like JPEG Ultra HDR and RAW + JPEG capture on compatible hardware.

There is an important limitation. The current Pixel 4 build is based on the initial Android 16 codebase, not the newer QPR2 release. That means some headline Android 16 features are missing, including the Material 3 Expressive design updates and Live Updates. The LineageOS team has started work on a QPR2-based LineageOS 23.2, but it is not available yet.

Installing LineageOS is not a simple update. It requires unlocking the bootloader, installing a custom recovery, and flashing the ROM manually. This process wipes the device and can break features like SafetyNet-dependent apps. It is not recommended for users who are uncomfortable with custom ROM installation or rely on locked-down banking or enterprise apps.

For users who are comfortable with those trade-offs, LineageOS 23 offers a practical way to keep an otherwise unsupported Pixel usable, secure, and relatively modern. Installation files and device-specific instructions are available on the LineageOS website.

Thank you for being a Ghacks reader. The post Google Pixel 4 Runs Android 16 With Official LineageOS 23 Support appeared first on gHacks Technology News.

WhatsApp is testing an updated Clear Chat experience on iPhone that makes it harder to delete the wrong things when cleaning up storage. The tool is not new, but the flow has been reworked to separate messages from saved items and media types, with clearer prompts and size estimates before anything is removed.

The change shows up when iOS users clear or delete a chat. Instead of a single confirmation prompt, WhatsApp now presents a two-step process that spells out what will be deleted and what can be kept. The aim is simple: free up space without accidentally wiping content that was intentionally saved, like starred messages or specific media categories.

The first screen adds quick actions for clearing messages in bulk. One option removes all messages. Another keeps starred messages while removing the rest. From there, a "Clear media files" option opens a second screen for more granular cleanup.

That second screen is the bigger upgrade. It breaks stored chat media into categories and lets users choose what to remove: documents, photos, videos, stickers, audio files, and voice messages. Each category shows a size estimate, and WhatsApp displays a running total so users can see how much storage will be recovered before committing.

Starred messages also get special handling. WhatsApp explicitly asks whether they should be kept or deleted during the clearing process. This matters for long-running chats where starring is used as a lightweight bookmark system for addresses, receipts, links, or important notes. The new flow reduces the chance of clearing a conversation "for space" and losing those saved items in the process.

The updated Clear Chat tool can be accessed from more than one place. Users can reach it from a chat's info screen, and it is also available from the main chats list, with the same detailed options in both entry points. That consistency helps because people tend to clean storage when they notice WhatsApp taking up space, and they may start from different parts of the app depending on habit.

What changed in practice

Older versions of WhatsApp's chat cleanup were easy to misread, especially if the goal was "delete media but keep messages," or "clear a chat but don't lose starred items." The revised prompts split the decision into smaller choices and attach storage numbers to those choices.

That combination does two things:

• It slows users down just enough to prevent accidental deletions.
• It makes storage cleanup more predictable by showing what each type of content costs in space.

For users trying to reclaim iPhone storage, the new screens also make it easier to target the actual storage hogs. In many chats, videos and documents dominate size, while messages contribute comparatively little.

How to use the new Clear Chat options on iPhone

This is currently in beta, so the exact labels can vary slightly, but the behavior follows the same pattern.

1. Open WhatsApp on iPhone.
2. Go to the chat you want to clean up.
3. Open the chat info screen (the chat header at the top), or start from the chats list entry for that conversation.
4. Choose the option to clear chat or manage chat storage.
5. On the first confirmation screen, pick one of the quick actions:
   • Delete all messages, or
   • Delete all messages except starred.
6. If your goal is storage cleanup, select Clear media files to open the second screen.
7. On the second screen, pick exactly what to remove:
   • Documents
   • Photos
   • Videos
   • Stickers
   • Audio
   • Voice messages
8. Review the size estimates per category and the total space to be freed.
9. Confirm the deletion.

If you rely on starred messages, treat the starred prompt as the critical checkpoint. That is where the new flow prevents the most common "cleanup regret" scenario.

Limitations and rollout status

The updated Clear Chat experience is rolling out gradually to users on the WhatsApp beta for iOS, specifically version 25.37.10.72, and it is not yet available to everyone. WhatsApp has not provided a public timeline for wider release.

Beta features can change before general rollout, including where the options appear and how the prompts are worded. Users who do not see the new screens are still on the older Clear Chat flow.

WhatsApp is also doing the right thing here by adding friction. Cleanup tools are destructive by nature, and the cost of a mistaken tap is usually higher than the benefit of a faster UI. Once media is deleted, recovery is not guaranteed, especially if the content is not saved elsewhere.

The new Clear Chat screens are a practical improvement for anyone who uses WhatsApp heavily, especially in large group chats where files accumulate silently over months.

 

Resources - https://wabetainfo.com/whatsapp-is-rolling-out-advanced-clear-storage-tool-with-selective-chat-cleanup-options-on-ios/

Thank you for being a Ghacks reader. The post WhatsApp for iOS Improves Clear Chat With More Control Over What Gets Deleted appeared first on gHacks Technology News.

Samsung is testing a revamped version of Bixby that integrates Perplexity AI, signaling that the assistant is not being sidelined in favor of Gemini after all. An early preview shows Bixby delivering more detailed, context-aware answers by tapping into Perplexity's search capabilities.

The preview surfaced via a screenshot shared on X, revealing a redesigned Bixby overlay responding to a weather query. Instead of a basic forecast, Bixby adds practical recommendations, such as suggesting a jacket, indicating deeper reasoning based on real-time information rather than static responses.

This functionality is currently limited to the One UI 8.5 beta and requires Bixby version 4.0.50.4. At the moment, access appears restricted to Galaxy S25, Galaxy S25 Plus, and Galaxy S25 Ultra devices enrolled in the beta. Samsung already uses Perplexity AI on its smart TVs, and this looks like the first step toward bringing similar search depth to Galaxy phones.

For users, the change means Bixby can handle more complex questions with fresher data, narrowing the gap with other AI assistants. The update also aligns Bixby's visuals with newer One UI design guidelines, making the assistant feel more integrated into the system rather than an afterthought.

Samsung has confirmed that One UI 8.5 will debut alongside the Galaxy S26, with a wider rollout to older Galaxy devices afterward. Until then, the Perplexity-powered Bixby remains a preview feature for beta testers, with broader availability expected once One UI 8.5 exits testing.

Thank you for being a Ghacks reader. The post Samsung Previews a Smarter Bixby Powered by Perplexity AI appeared first on gHacks Technology News.

Sony is offering a time-limited perk for PlayStation 5 owners: three free months of Apple Music when accessed through the console. The promotion has a retail value of about $30, but not everyone will qualify.

The offer is available only through the Apple Music app on PS5 and applies to new subscribers and some "qualified returning" users. Sony and Apple have not published exact eligibility rules for returning accounts, which suggests the trial is restricted to users who have not had an active Apple Music subscription for a certain period.

To claim the offer, PS5 owners need to download and open the Apple Music app on the console and sign in with an Apple ID. If the account qualifies, the three-month trial is applied automatically. The promotion runs until March 18, 2026, and is exclusive to PS5-other Apple Music platforms do not trigger the deal.

The free trial provides full access to Apple Music's catalog and PS5 integration, including background playback while gaming. After the trial ends, the subscription renews at the standard monthly rate unless canceled.

This is not the first time Sony has partnered with Apple on console-specific trials, but the catch remains the same: eligibility is limited, and existing subscribers should not assume the offer will apply to their accounts. Users who want the trial should verify eligibility directly from the PS5 app before the March deadline.

Are you using Apple Music or prefer Spotify?

Thank you for being a Ghacks reader. The post PS5 Owners Can Get Apple Music Free for Three Months, But With Limits appeared first on gHacks Technology News.

Windows' built-in copy function works well enough for small files. Problems start when transfers involve tens or hundreds of gigabytes, or thousands of files. At that point, File Explorer often slows to a crawl, stalls on "Calculating time remaining," or fails partway through with little clarity on what actually copied.

The issue is not storage speed. It is how File Explorer handles large transfers. Before copying begins, Windows tries to enumerate every file and estimate total size and time. On large directories, this pre-calculation alone can take minutes or longer, and the estimates remain unreliable throughout the process.

Error handling is another weak point. If one file is locked or unreadable, Explorer frequently pauses the entire operation and waits for user input. In some cases, the transfer aborts, leaving a partially copied directory with no built-in way to verify what succeeded. Resume support exists, but re-verification is slow and inefficient, especially across external or network drives.

Explorer also assumes success equals integrity. It does not verify copied data with checksums. For backups, archives, or large media files, this means silent corruption can go unnoticed until the file is opened later.

For large or critical transfers, command-line tools are more reliable. Windows includes Robocopy (Robust File Copy), which is designed for bulk data movement and directory mirroring. It avoids GUI overhead, supports retries, logs every action, and can resume interrupted transfers cleanly.

Robocopy also allows multithreaded copying, making it far more efficient when dealing with thousands of small files. Options exist to control retry counts, wait times, and behavior when files are locked. For backups or migrations, it can mirror directories exactly, reducing human error.

A simple example looks like this:

robocopy D:\Source E:\Backup /MIR /R:3 /W:5 /MT:8

This mirrors one folder to another, retries failed files three times, waits five seconds between retries, and uses multiple threads.

Robocopy is not risk-free. A wrong flag, especially with mirroring, can delete data at the destination. Users should test commands on non-critical folders first and read output logs carefully.

For occasional small transfers, File Explorer remains fine. For professionals, content creators, or anyone regularly moving large datasets, relying on Explorer alone is asking for delays and uncertainty. Command-line tools trade convenience for control, but they finish transfers predictably and leave an audit trail when something goes wrong.

Thank you for being a Ghacks reader. The post Why Windows File Copy Struggles With Large Files, and What Works Better appeared first on gHacks Technology News.

Players logging into World of Warcraft today will find free holiday rewards waiting under the Feast of Winter Veil tree. Blizzard's annual Winter Veil tradition returns with three exclusive items that can be claimed right now, but only for a limited time.

The 2025 Winter Veil gifts include two toys and one battle pet: the Rolling Snowball, the Jubilant Snowman Costume, and the Tiny Snow Buddy. These rewards are available once per character and can be collected by visiting the Feast of Winter Veil tree in the capital cities.

Winter Veil gifts are only available starting December 25 and must be claimed before the event ends on January 2. Players need to interact with the presents under the tree, complete the short quests they provide, and then open the gifts to permanently add the items to their collections.

This annual giveaway has been part of World of Warcraft since 2007. Each year's gifts are new, while older holiday items typically become much harder to obtain after the event ends. Missing the initial window does not make the rewards impossible to get, but it significantly reduces the odds.

Gifts from past Winter Veil events can sometimes drop from the Stolen Present, which is rewarded for completing the Abominable Greench daily quest during the holiday. These drops are rare and tied to a Feat of Strength achievement.

Players who want the items with certainty should log in before January 2 and collect the gifts on every character they care about, as each character can claim the rewards independently.

Thank you for being a Ghacks reader. The post World of Warcraft Players Can Claim Free Winter Veil Gifts Until January 2 appeared first on gHacks Technology News.

OpenAI has begun rolling out a subtle interface change in ChatGPT that alters how certain responses are displayed. The update introduces "formatting blocks," which present content like emails or documents inside a task-style layout instead of the standard chat bubble.

The change appears when ChatGPT detects writing-oriented prompts, such as drafting emails or longer text. Instead of returning a plain message, the output is shown in a formatted document area. Selecting text inside these blocks reveals a small editor toolbar with common formatting options, similar to what users see in email or document apps.

This does not change how ChatGPT generates text. The update affects presentation only. The goal is to make editing drafts easier without copying content into another app, especially for tasks that resemble document creation rather than conversation.

Formatting blocks currently appear in GPT's newer rich-text responses and are rolling out gradually. Not all users see them yet, and support varies depending on the task type. OpenAI has not published a full list of triggers or a timeline for broader availability.

There is no dedicated toggle to disable formatting blocks at the moment. Users who prefer plain chat output can usually avoid the new layout by asking ChatGPT to respond in plain text or by framing prompts explicitly as conversational replies rather than documents. Switching to shorter prompts or requesting "plain text only" also tends to bypass the formatted view.

The update nudges ChatGPT closer to task-oriented tools like email editors and note apps, at least for writing-heavy workflows. For users who treat GPT primarily as a conversational assistant, the change may feel unnecessary, but it does not affect prompts that do not trigger document-style output.

Thank you for being a Ghacks reader. The post ChatGPT’s New Formatting Blocks Change How Writing Tasks Look appeared first on gHacks Technology News.

A new native NVMe storage driver introduced in Windows Server 2025 can significantly improve SSD performance, and users have found a way to enable it on Windows 11 as well. The method relies on manually toggling feature flags in the Windows registry. While Microsoft does not officially support this on consumer versions of Windows, early testing shows measurable gains under certain conditions.

The driver is designed to reduce CPU overhead and improve IOPS performance for NVMe SSDs. According to Microsoft, the Server 2025 implementation can deliver up to 80 percent higher IOPS and cut CPU usage by roughly 45 percent. Those figures apply to server workloads and do not translate directly to desktop usage.

Community testing on Windows 11 version 25H2 shows more modest results. Real-world benchmarks reported by third-party testers indicate performance improvements closer to 10-15 percent. That still represents a meaningful gain for storage-heavy workloads, but it falls well short of Microsoft's server-side claims.

The feature is not exposed through standard Windows settings. Enabling it requires editing the registry to activate internal feature overrides that switch Windows 11 to the newer NVMe driver. After activation, NVMe drives appear under "Storage Media" instead of "Devices" in Device Manager, confirming the driver change.

How the registry change works

Warning: Editing the registry can cause system instability. Back up the registry or create a restore point before proceeding.

1. Press Win + R, type regedit, and press Enter.
2. Navigate to:HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Policies\Microsoft\FeatureManagement\Overrides
3. Create the following DWORD (32-bit) values and set each to 1:
   • 735209102
   • 1853569164
   • 156965516
4. Restart the system.

After rebooting, Windows 11 loads the newer NVMe driver if the SSD and system configuration are compatible.

Known issues and rollback

Some users report compatibility problems with SSD management utilities, including Samsung Magician, after enabling the driver. Device monitoring features or firmware tools may fail to detect drives correctly.

To roll back the change, delete the added DWORD entries from the same registry location and restart the system. Windows will revert to the default NVMe driver automatically.

This tweak is best viewed as an experimental optimization rather than a guaranteed upgrade. Gains vary by workload, and the lack of official support means updates or future builds could disable the behavior without notice.

Thank you for being a Ghacks reader. The post This Registry Hack Unlocks a Faster NVMe Driver in Windows 11 appeared first on gHacks Technology News.

Google is rolling out a long-requested feature that allows users to change their primary @gmail.com address without creating a new Google account. The change gives users more control over a core part of their identity that was previously locked in for the lifetime of an account.

Until now, changing a Gmail username meant starting over with a new account or relying on aliases and secondary addresses. Google Accounts could host multiple emails, but the original Gmail address always remained the primary identifier. Google is now relaxing that limitation for accounts that already use a Gmail address.

The option is rolling out gradually and appears to be enabled region by region. Early documentation surfaced in India first, suggesting a phased deployment. Users who have access can find the setting under Google Account ? Personal Info ? Email.

The feature is limited to Gmail addresses only. Both the old and new addresses must end in @gmail.com, and the change cannot be used to switch to a custom domain or another provider. Importantly, all existing data stays intact. Emails, Google Drive files, Photos, YouTube history, and other account data remain unchanged after the switch.

Google keeps the old Gmail address active as a secondary address. Emails sent to the previous address will still be delivered, and users can sign in using either address. This reduces the risk of missing messages or being locked out of third-party services tied to the old email.

There are strict limits. Each account can generate a new Gmail address only three times. After a change, users must wait 12 months before creating another new address. Google also warns Chromebook users to back up local files first, as removing and re-adding the account may wipe the device's home directory.

Some friction is expected. Certain apps may reset preferences, and older content such as calendar events or shared documents may continue to display the original email address.

The rollout finally brings Gmail closer to the flexibility long offered by other email providers, though the limits make it a feature best used carefully.

Thank you for being a Ghacks reader. The post Google Now Lets Users Change Their Gmail Address Without a New Account appeared first on gHacks Technology News.

Microsoft rolled out several Copilot Studio updates in November 2025, largely shaped by announcements at Microsoft Ignite. The changes reinforce a clear direction: Copilot Studio is evolving from a workflow automation tool into a platform for building and governing AI agents inside organizations.

One of the most notable updates is the general availability of GPT-5 Chat in both the US and the EU. This removes a regional limitation that previously affected enterprise deployments and allows organizations to standardize agent behavior across markets.

Another key change is the introduction of human-in-the-loop (HITL) controls, now available in preview. HITL allows organizations to require human review or approval at specific stages of an agent's execution. This is aimed at reducing risk in scenarios where AI agents interact with sensitive data, trigger downstream actions, or make decisions that require accountability.

Microsoft is positioning these controls as a governance layer rather than a speed bump. Agents can still operate autonomously within defined boundaries, but escalation paths and checkpoints are built into the system by design.

Taken together, the November updates highlight a broader shift in how Copilot Studio is expected to be used. Instead of automating isolated steps, organizations are being encouraged to design systems where AI agents manage complexity, while humans provide judgment, validation, and oversight.

There are no major consumer-facing changes in this release. The focus remains squarely on enterprise use cases, compliance requirements, and large-scale operational workflows.

Thank you for being a Ghacks reader. The post Microsoft Expands Copilot Studio With AI Agent and Governance Features appeared first on gHacks Technology News.

Google has quietly expanded and restructured its paid AI offerings, and by December 2025 the differences between the free tier, Google AI Pro, and Google AI Ultra are no longer subtle. The plans now differ sharply in daily limits, model access, context size, and which products are unlocked.

At the center is the Gemini app. Free users get general access to Gemini 3 Flash, limited access to Thinking and Pro models, a 32,000-token context window, and capped features like Audio Overviews and Deep Research. Google has recently expanded free access in the US, but strict daily limits still apply.

Google AI Pro replaces the older "AI Premium" branding and costs $19.99 per month in the US. The plan significantly raises Gemini usage caps, including up to 100 prompts per day with Thinking and Pro models, 20 Deep Research reports per day, and a 1-million-token context window. Image generation limits jump dramatically, scheduled actions are enabled, and spreadsheet and code analysis become available.

AI Pro also unlocks Gemini features across Google Workspace apps such as Gmail, Docs, Sheets, Slides, and Meet. Google Vids is included as well, bringing AI-assisted video creation tools aimed at work and education use cases. Subscribers also receive higher limits in Google Search's AI Mode, where Deep Search can generate long, cited reports after running hundreds of background queries.

Beyond Gemini, AI Pro bundles access to NotebookLM Pro features, including higher Audio Overview limits, more notebooks and sources, advanced chat settings, and analytics. The plan includes 2 TB of Google One storage, which already accounts for half the monthly cost if purchased separately.

Google AI Ultra sits at the opposite end of the spectrum. Priced at $249.99 per month, it raises nearly every cap to what Google labels as "Highest" access. Gemini prompts increase to 500 per day, Deep Research jumps to 200 reports daily, and Deep Think enables large reasoning tasks with a 192,000-token context window. Image and video generation limits expand again, and agent-based workflows can run in parallel.

Ultra subscribers also receive expanded Google Home Premium Advanced features, the highest Google Photos AI creation limits, and dramatically increased NotebookLM quotas, including video overviews, quizzes, flashcards, and report generation. Storage increases to 30 TB, and YouTube Premium is bundled in. New AI Credits are included for experimental tools like Whisk and Flow, with additional credits available for purchase.

The gap between Pro and Ultra is no longer about casual use versus power use. AI Pro targets heavy individual users and professionals who rely on Gemini daily. AI Ultra is designed for intensive, multi-agent workflows, large research output, and media generation at scale.

Thank you for being a Ghacks reader. The post What You Actually Get With Google AI Pro and AI Ultra in December 2025 appeared first on gHacks Technology News.

Apple has released iOS 26.2, and unlike some system updates that focus on under-the-hood changes, this one delivers visible improvements across several built-in iPhone apps. Six apps in particular receive new features that users are likely to notice immediately in daily use.

Reminders gain a long-requested capability. Tasks with a specific due time can now be marked as Urgent, triggering an alarm when the reminder is due. The alert behaves like a standard alarm, including a 9-minute snooze option, and also appears as a Live Activity. This turns time-sensitive reminders into something closer to calendar alarms, closing a long-standing gap in the app.

Apple Podcasts receives multiple enhancements aimed at discovery and navigation. The most notable is automatic chapter generation. Apple uses podcast transcripts combined with its on-device AI models to generate chapters even for shows that do not provide them. This makes skipping and revisiting segments easier, especially for longer episodes. Podcasts also surfaces links to episodes, shows, and media mentioned during playback.

Freeform adds support for tables. The app's open-canvas design already allowed text, images, links, and files, but structured data was missing. Tables bring Freeform closer to Notes in flexibility, particularly for planning, comparison, and collaborative workspaces.

Apple Games, which debuted earlier in iOS 26, gets several refinements. Library filters make it easier to sort installed games, controller support is expanded, and real-time challenge updates improve visibility for multiplayer activity. These changes focus on usability rather than introducing new gaming features.

Apple Music finally allows lyrics to be available offline. Previously, lyrics required an active internet connection even for downloaded songs. With iOS 26.2, lyrics can appear when offline, such as during flights or in areas with poor reception. Apple has not clarified exactly when lyrics are downloaded, but songs saved to the library appear to benefit.

Apple News sees interface changes that make existing features easier to find. New quick links at the top of the Today tab surface sections like Puzzles, Sports, Politics, and Food. The navigation bar has also been reorganized, moving some content out of Search and into a dedicated Following tab.

iOS 26.2 does not overhaul the system, but it meaningfully improves several apps people already use every day.

Thank you for being a Ghacks reader. The post iOS 26.2 Adds New Features to Reminders, Music, Podcasts, and More appeared first on gHacks Technology News.

Google has started rolling out ChromeOS 143, and for most regular users, the update passes without any visible changes. Like several recent releases, this version focuses almost entirely on enterprise and education environments rather than consumer-facing features.

The most significant changes are aimed at IT administrators. ChromeOS 143 improves management of USB-connected printers, an area that previously lagged behind network printer support. Admins can now control locally attached printers using vendor and product IDs, making it easier to restrict which peripherals are allowed in managed environments.

Google has also added more granular print controls. Administrators can predefine or restrict settings such as paper size, color mode, duplex printing, and DPI on a per-printer basis. These controls are designed for organizations where consistent print output is required, such as healthcare, logistics, or education.

On the application side, Google Vids is now pre-installed for managed work and school accounts. The AI-assisted video creation app appears automatically in the ChromeOS launcher for those users. For personal Chromebook owners, this change has no practical impact.

There is one minor user-facing tweak. ChromeOS 143 adds a Kana/Romaji toggle directly to the Japanese virtual keyboard. This is mainly useful for kiosk and shared-device setups where access to full system settings is restricted.

Google's release notes also point ahead to upcoming badge-based authentication features planned for ChromeOS 145, reinforcing the platform's current direction toward frontline, enterprise, and education use cases.

For consumers, ChromeOS 143 does not change how Chromebooks look or behave day to day. The update continues a pattern where ChromeOS development prioritizes backend management and organizational tooling over new productivity or interface features.

Is Google Chrome your main browser? Or do you prefer others?

Thank you for being a Ghacks reader. The post ChromeOS 143 Rolled Out Quietly With Changes Most Users Won’t Notice appeared first on gHacks Technology News.

The most serious cyber risks consumers face in 2026 are less about technical break-ins and more about manipulation. Criminals increasingly rely on realistic AI-generated media and social engineering to pressure people into acting quickly, often before they have time to verify what is happening.

Law enforcement agencies, including the Federal Bureau of Investigation, have warned that scammers are already using altered or fabricated audio and video as "proof of life" in extortion and virtual kidnapping schemes. Photos and clips pulled from social media are remixed to create convincing scenarios designed to trigger panic and urgency.

This shift is enabled by generative AI. The technology does not create new crimes, but it lowers the cost and effort required to run old ones at scale. The Federal Trade Commission has issued similar warnings, noting that voice cloning and realistic media are being used to make fraud harder to detect.

In practice, most successful scams exploit one of three weaknesses: urgency, account access, or oversharing. Deepfake pressure scams rely on emotional urgency. Account takeovers target email, cloud, or mobile carrier accounts to cascade into wider access. Oversharing increasingly happens through AI chatbots, where users paste sensitive details assuming privacy that may not fully exist.

Public networks remain a risk multiplier. Untrusted Wi-Fi environments still expose users to interception and credential theft, reinforcing long-standing guidance to avoid sensitive actions on public networks or use encrypted connections where possible.

Defensive advice for 2026 centers on reducing impact rather than spotting every fake. Strong account protection, including multi-factor authentication and passkeys, limits damage even if credentials are exposed. Trimming public personal information reduces what attackers can weaponize. Layered protections such as malicious-site blocking, backups, and verification rules for urgent requests remove leverage from attackers.

AI-driven scams succeed by speeding people up. Effective defenses slow things down, add verification steps, and reduce how much damage a single mistake can cause.

Thank you for being a Ghacks reader. The post Consumer Cyber Risks in 2026 Focus on AI-Driven Scams, Not Hacks appeared first on gHacks Technology News.

The postmarketOS project has released version 25.12, bringing incremental but meaningful improvements to its Linux distribution for phones and tablets. The update does not turn postmarketOS into a drop-in replacement for Android or iOS, but it continues to narrow the gap with better hardware support, system upgrades, and usability improvements.

The release moves the distribution to Alpine Linux 3.23 and switches to version 3 of the Alpine Package Manager. One practical benefit is safer updates: packages are now downloaded fully before installation, reducing the risk of a broken system when updating over unstable connections.

User interface components also see updates across the board. GNOME has moved to version 49, though the mobile-optimized variant remains on GNOME 48 for now. KDE Plasma Mobile has been updated to 6.5.3, improving lock screen performance, home screen behavior, and Waydroid integration for running Android apps. Phosh and Sxmo have also been refreshed to their latest releases.

The project's customized Firefox build has been refined as well. All mobile-specific tweaks can now be controlled from a dedicated settings page, including address bar placement, tab behavior, and menu interactions. Popup reliability has also been improved, addressing a common pain point on touch devices.

Hardware support grows modestly in this release. The Lenovo ThinkSmart View, originally a desktop video conferencing device, is now supported as a community device with working touch input, Wi-Fi, and 3D acceleration. Audio, camera, Bluetooth, and ambient light sensors remain incomplete, highlighting the experimental nature of many ports.

For Linux enthusiasts interested in mobile platforms, postmarketOS 25.12 continues the project's steady progress. It remains best suited for experimentation, development, or secondary devices rather than daily-driver use, but each release reduces friction and expands what is realistically usable.

Thank you for being a Ghacks reader. The post Linux postmarketOS 25.12 Adds More Devices and UI Updates appeared first on gHacks Technology News.

Google is testing an expansion of one of NotebookLM's most useful features. A new mode spotted in development suggests that Audio Overviews may soon support longer, lecture-style audio summaries generated from user-provided sources.

The information was surfaced by Android Police, based on findings from a leaker who identified a new "Lecture" format inside NotebookLM's Audio Overviews. The mode appears alongside existing options such as Brief, Deep Dive, Critique, and Debate.

The Lecture format is designed to generate a single, continuous audio explanation of a topic, structured more like a class session than a summary. Users would be able to choose the lecture length, with options labeled Short, Default, and Long. According to the demo, a Long lecture can run for roughly 30 minutes and is delivered by a single AI host.

The feature also appears to include a language selector, allowing users to choose the spoken language for the lecture. Google has not confirmed which languages would be supported or whether this option would be limited to specific regions or account tiers.

If released, the Lecture mode would push NotebookLM further toward passive learning and background consumption, making it easier to absorb long documents, research material, or notes without actively reading. It would also reduce the need to manually break sources into smaller summaries.

There is no official timeline for rollout, and Google has not confirmed whether the Lecture mode will be available to all users or restricted to higher tiers. Until announced, the feature remains in testing and may still change or be removed.

Thank you for being a Ghacks reader. The post NotebookLM Is Testing a New “Lecture” Mode for Audio Overviews appeared first on gHacks Technology News.

Google has released Android 16 QPR3 Beta 1.1, a small but important update that fixes a bug causing certain apps to crash on launch. The update arrives less than a week after the first QPR3 beta and targets an issue that quickly frustrated Pixel users enrolled in the beta program.

The problem appeared shortly after users installed Android 16 QPR3 Beta 1. Reports pointed to repeated startup crashes affecting apps such as Microsoft Teams, OneDrive, and several banking apps, making daily use difficult on affected devices.

Beta 1.1 is now available for all eligible Pixel phones enrolled in the Android Beta Program, starting with the Pixel 6 series and extending through newer models, including foldables. Most devices receive build CP11.251114.007, while the Pixel 7a gets a separate build variant.

Google's release notes for the update are brief, listing a fix for apps that crash during startup. The OTA package is small, around 60MB on devices like the Pixel 10 Pro XL, and should automatically appear for users already on the beta channel.

Early feedback suggests the update resolves crashes for most banking apps, restoring basic usability. Some Microsoft apps, including Teams and Outlook, are still reported to crash on launch, and clearing app or system cache does not appear to help in those cases.

Google also warns users who plan to leave the Android Beta Program. Installing QPR3 Beta 1.1 and then exiting the program may result in data corruption. Users who intend to opt out are advised to skip this update or wait for QPR3 Beta 2 before leaving the beta track.

For users staying on the beta, the update is recommended, as it addresses a widespread and disruptive issue without introducing major changes.

Thank you for being a Ghacks reader. The post Android 16 QPR3 Beta 1.1 Fixes App Crash Issue on Pixel Devices appeared first on gHacks Technology News.

Google is testing a new Android feature designed to anticipate what users want to do next and surface apps or actions automatically. The feature, called Contextual Suggestions, has been spotted in beta and is tied to Google Play Services rather than a full Android version update.

The discovery comes from Android Authority, which found references to Contextual Suggestions in Google Play Services version 25.49.32. The feature aims to proactively present relevant actions based on context, such as showing navigation and music controls when heading to a car, or surfacing playlists at the gym.

Other examples include automatically suggesting media casting when a TV is detected, or bringing up apps tied to a routine without manual input. The system appears designed to reduce friction by predicting intent rather than waiting for explicit commands.

According to settings strings uncovered in the beta, Contextual Suggestions are processed locally on the device. Google states that the data used is encrypted and automatically deleted after 60 days. Users can also manually clear the data at any time.

The feature is optional. Android includes a toggle that allows Contextual Suggestions to be disabled completely, preventing the system from generating or acting on predictions. This aligns with Google's recent approach of shipping AI-driven automation as opt-out rather than mandatory.

Despite local processing claims, the feature still represents a shift toward deeper behavioral inference on Android devices. Predictive systems rely on usage patterns, location signals, and activity context, which may concern users who prefer explicit control over automation.

Contextual Suggestions are currently limited to beta builds and are not yet widely available. Google has not announced a rollout timeline or confirmed which Android versions will support the feature.

Thank you for being a Ghacks reader. The post Google Is Testing an Android Feature That Predicts What You Do Next appeared first on gHacks Technology News.