IT News Review by Control F5 Software: Roblox implements age verification with artificial intelligence for child protection

Roblox implements age verification with artificial intelligence for child protection

Roblox introduces a new mandatory age verification system, based on artificial intelligence and identity documents, in response to lawsuits accusing the platform of allowing abuse against minors. Users will confirm their age either by uploading an identity document or by taking a photo of their face, based on which an AI system will estimate their age.

Depending on the result, each user will be placed into an age category, and conversations will only be allowed between users of similar ages. For example, a child estimated to be 12 years old will be able to talk to users up to 15 years old, but not with those aged 16 or older.

The feature becomes mandatory in December in Australia, New Zealand, and the Netherlands and will be rolled out globally at the beginning of next year. The decision comes amid lawsuits filed by the attorneys general of Kentucky and Louisiana for harming children, as well as a criminal investigation opened in Florida. Roblox already uses parental controls and AI moderation for text and voice, but the new update will massively increase the number of users verifying their age on the platform.

CrowdStrike fires an employee suspected of helping hackers

The cybersecurity giant CrowdStrike has confirmed the firing of an employee suspected of providing internal information to a notorious group of hackers. The group Scattered Lapsus$ Hunters published screenshots suggesting internal access to CrowdStrike's systems, including Okta dashboards used by employees to access internal applications. The hackers claimed they gained access through a recent breach at Gainsight, a customer relationship management company.

CrowdStrike, however, rejects the hackers' claims and states that the company's systems were not compromised. According to official statements, the insider's access was revoked after it was found that they were sharing screenshots of their own computer externally. Clients were reportedly protected throughout the incident, and the case has been referred to the relevant authorities.

Scattered Lapsus$ Hunters is a collective made up of several groups, including ShinyHunters, Scattered Spider, and Lapsus$, known for using social engineering to trick employees and gain access to sensitive systems or databases.

Meta accused of hiding evidence about the negative impact of social networks

U.S. court documents suggest that Meta may have hidden evidence regarding the negative effects of its social media platforms on users, especially on young people. The context indicates serious accusations about how the company has managed information related to the impact of Facebook and Instagram.

These new accusations add to a series of legal and regulatory controversies that Meta has faced in recent years, including concerns related to the mental health of teenagers and aggressive data collection practices.

Snapchat introduces age verification in Australia ahead of social media ban for minors

Snapchat has begun asking teenagers in Australia to verify their age, just a few weeks before the country implements a radical law banning access to social media for those under 16.

Starting December 10, platforms like Facebook, Instagram, and TikTok will be required to remove users under 16 or face significant fines. Age verification can be done through an Australian bank account, a government-issued ID, or a photo of the face, based on which a third party will estimate the age range.

User accounts under 16 will be blocked from December 10. Snapchat says it does not agree with being included in this ban but will comply with the law, as it does with all local regulations. The company warns, however, that disconnecting teenagers from friends and family does not necessarily make them safer and could push them towards less safe and less private apps. So far, ten platforms, including Discord, WhatsApp, and Pinterest, have managed to avoid being included in this historic legislation.

Children in China use bots and engagement tricks on social media

Young people in China are developing ingenious methods to "trick" the algorithms of social media platforms like Xiaohongshu (Little Red Book).

Among the tactics are the use of automated bots and various engagement tricks to increase the visibility of content or even to help them complete school tasks. The phenomenon shows how quickly children and teenagers become adept at the digital mechanisms of platforms, even in a context where the state imposes strict limits on internet use by minors.

Major music labels sign with a new AI streaming service

Several major record labels have signed agreements with a new AI-based streaming service, marking a turning point for the music industry. Although the full details of the service and contracts are not yet public, the direction is clear: AI is becoming an integral part of how music is distributed and consumed.

The new agreements suggest an increasing openness of the industry to AI technologies, after years of reluctance towards digital innovations. AI promises much more personalized recommendations, rapid discovery of new content, and possibly new ways to create and remix music.

Nokia announces a new strategy and major operational restructuring

Nokia has presented its new strategy and a broad evolution of its operational model, setting long-term financial goals and changes in the leadership team. The company will reorganize into two main segments: Network Infrastructure and Mobile Infrastructure, with the ambition to lead the transformation of AI-based networks and to leverage the so-called "AI supercycle".

Nokia's goal is to increase comparable annual operating profit to €2.7 – 3.2 billion by 2028, compared to the €2 billion generated in the last 12 months.

The Network Infrastructure segment is positioned to benefit from the global boom in AI data center construction, while the new Mobile Infrastructure targets AI-native networks and the transition to 6G. The company has also identified several units considered non-essential for its future strategy, which it will move to a separate segment called Portfolio Businesses, with net sales of approximately €0.9 billion.

The strategy focuses on accelerating growth in AI and cloud, leading the next era of mobile connectivity, and achieving sustainable outcomes.

Spotify tests a playlist import feature

Spotify is testing a new feature that allows users to import playlists from other streaming services. Essentially, you can take your music collections with you without having to rebuild them manually.

This new feature could eliminate one of the biggest barriers to switching streaming platforms: the fear of losing playlists built over years. In a context of intense competition among streaming services, a quick and simple import feature could convince more users to try alternatives to their current platform.

Gmail offers an option to block data use for AI training

Google has introduced an option in Gmail that allows users to refuse the use of their data for training artificial intelligence models. The measure comes amid growing concerns about privacy and how tech companies use personal information to improve their AI systems.

The new option gives users more control over their data, although the exact way it is implemented and the actual level of protection offered remain hotly debated in the privacy community. The trend is clear: in the AI era, users are demanding more transparency and control over how their data is used.

Notepad receives table support in Windows

Microsoft adds table support to the classic Notepad in Windows, transforming one of the simplest applications in the system into a more versatile tool. Users will be able to create and manage structured data in the form of tables directly in Notepad.

The update extends the utility of the application beyond simple text editing and makes it more useful for organizing information, creating structured lists, or managing small data sets. It is a new piece in a series of improvements made to Notepad in recent years, through which Microsoft seeks to modernize the application without compromising the simplicity that made it so popular.

Microsoft PowerToys receives the Advanced Paste feature with on-device AI

Microsoft has launched a new functionality called Advanced Paste in the PowerToys suite, which uses artificial intelligence running locally on the device. The goal: to make copy and paste operations much smarter.

With Advanced Paste, users can transform content from the clipboard in one step: from automatic text formatting to translations or extracting specific information, all processed locally without sending data to the cloud.

The fact that the AI runs on-device is a significant privacy plus, especially for sensitive data. PowerToys thus continues to be a highly appreciated set of tools for advanced Windows users, constantly expanding the default capabilities of the system.

OpenAI launches group chats in ChatGPT

OpenAI has introduced the group chat feature in ChatGPT, allowing multiple people to collaborate in the same conversation with the AI assistant. ChatGPT thus transitions from a "single-user tool" to a collaborative platform.

Teams can now discuss ideas, brainstorm, or work together on projects, all while having the support of AI. The feature opens up new use cases, from educational group sessions to creative workshops or project planning. It is yet another step in the evolution of AI assistants from simple personal tools to intelligent collaborative spaces.

Perplexity launches the AI browser Comet on Android

Perplexity has launched Comet, a new AI browser for Android that integrates search and AI-generated answers directly into the browsing experience.

Instead of switching between a browser and a chatbot, Comet provides direct answers to questions, summaries of web pages, and additional context while you browse. It is a new approach to how we explore the internet: AI is no longer a separate tool but an active guide in the browser. The launch on Android brings it to a wide audience of mobile users who want a faster and smarter browsing experience.

Google's Gemini arrives on Android Auto

Google has integrated its AI assistant, Gemini, into Android Auto, bringing advanced artificial intelligence capabilities directly into the car. Drivers can talk to Gemini to plan routes, find locations, or get real-time information without taking their eyes off the road.

The integration is part of Google's strategy to expand the Gemini ecosystem beyond phones and laptops and bring it into contexts where intelligent voice assistance can be particularly useful. Gemini promises a more natural and conversational interaction than traditional voice assistants, helping drivers manage complex tasks while staying focused on driving.

Nvidia releases a patch for performance issues in Windows 11

Nvidia has released a hotfix driver to address serious performance issues in games that arose after the October 2025 update of Windows 11. The Microsoft update KB5066835 caused massive drops in frame rates in certain games for Nvidia graphics card users.

Independent tests have shown improvements of up to 50% after installing the hotfix, with cases of increases from 149 FPS to over 200 FPS in games like Assassin’s Creed Shadows. The Hotfix Display driver version 581.94 fixes a communication issue between the game, Desktop Window Manager, and the PC hardware, related to changes in the Multi-Plane Overlay system in Windows 11.

Nvidia recommends installing the hotfix only for users who are actually experiencing these issues, as the driver is in beta status and has undergone limited testing. Users with AMD and Intel cards remain affected for now, without a similar solution.

Adobe buys Semrush for $1.9 billion

Adobe has announced the acquisition of the digital marketing platform Semrush for approximately $1.9 billion in cash, at $12 per share. This is Adobe's first major acquisition after the failure of the $20 billion Figma deal in 2022.

Semrush, one of the top platforms in SEO and brand visibility, will be integrated into Adobe's Digital Experience portfolio. The goal is to help marketers better understand how their brands appear in traditional search results, in AI-generated responses, and on the web in general.

The acquisition comes at a time when the way people discover information online is fundamentally changing: traffic from generative AI sources to retail sites in the U.S. has increased by 1,200% year-over-year as of October 2025. Adobe plans to connect Semrush's SEO and GEO capabilities with products like Adobe Experience Manager and Adobe Analytics.

The deal is expected to close in the first half of 2026, subject to regulatory approvals and shareholder votes. Semrush also owns media properties like Search Engine Land, raising questions about how Adobe will manage these editorial assets.

WhatsApp introduces a feature similar to Instagram Notes

WhatsApp is launching its own version of short status updates, similar to Instagram Notes, allowing users to post short text messages visible to others. The feature uses the "About" section, which was actually WhatsApp's first feature before the app focused on private and encrypted messaging.

WhatsApp suggests that people can use this space to share life updates, start conversations, or explain why they are unavailable. With the relaunch, the About status becomes much more visible, appearing at the top of individual chats and on user profiles.

Like Instagram Notes, the About status will disappear by default after one day, but users can adjust the time to make it disappear faster or remain for more than 24 hours. For now, the feature is not as complex as Instagram Notes, which accepts short looping videos and music integration, but if it is widely adopted, we may see such improvements in the future.

YouTube tests private messaging in the app again

After discontinuing the private messaging feature in 2019, YouTube is testing this option again for a limited group of users. Those included in the test can share videos directly in the mobile app, whether they are long clips, Shorts, or livestreams.

By pressing the Share button, users can open a full-screen chat and start a private one-on-one or group conversation. Friends can respond with their own videos, text messages, or emojis.

The test is available for users aged 18 and older in Ireland and Poland. It is a significant change, given that currently, the only way to share YouTube videos is through SMS, email, WhatsApp, iMessage, Instagram DM, or other messaging platforms. YouTube states that in-app messaging has been a "top request" from users and includes standard safety measures, such as mutual invitations before starting a chat, the ability to unsend messages, block others, and report conversations.

The best guide for detecting AI writing comes from Wikipedia

Wikipedia has created a detailed guide for identifying texts generated by artificial intelligence, based on the experience of editors who face thousands of attempts to introduce such content on the platform. Unlike automated tools, which are often ineffective, the guide focuses on expression patterns and specific stylistic habits of AI models.

Among the signals are overly pompous language, a generic tone, excessive use of lists and emojis, as well as the presence of prompts or instructions left inadvertently in the text. AI models tend to exaggerate the importance of subjects, using phrases like "a crucial moment" or "a broader movement," and sometimes treat minor appearances in the media as evidence of "notability," similar to a personal biography.

The guide also draws attention to typical formulations in closing sentences, where models declare that an event "highlights the significance" or "reflects the ongoing relevance" of an idea. Wikipedia emphasizes that this is not a list of banned words, but rather a catalog of very common patterns observed in thousands of examples of AI-generated text.

Google must double its AI serving capacity every six months

The head of AI infrastructure at Google has told employees that the company must double its AI serving capacity every six months to keep up with demand. In a presentation in November 2025, Amin Vahdat, Google Cloud vice president, stated: "We now need to double every 6 months... the next increase of 1000x in 4-5 years."

He explained that the infrastructure for AI is both the most critical and the most expensive component of the AI race. The presentation came a week after Alphabet reported financial results that exceeded expectations for the third quarter and raised its capital expenditure forecast again to $91-93 billion, anticipating a "significant increase" in 2026.

CEO Sundar Pichai described 2026 as "intense," citing fierce competition in AI and the pressure to meet the growing demand for cloud and compute. He was also asked about the risk of a potential "AI bubble," an increasingly discussed topic in Silicon Valley and Wall Street amid plans for trillions of dollars in investments in the coming years.

A simple vulnerability in WhatsApp exposed 3.5 billion phone numbers

Researchers from the University of Vienna managed to extract phone numbers for 3.5 billion WhatsApp users, systematically using the contact discovery feature. Essentially, they tested number after number and checked which of them were associated with WhatsApp accounts.

The technique allowed them to obtain profile pictures for 57% of accounts and profile text for 29%. The researchers checked approximately 100 million numbers per hour, using the web version of WhatsApp. The team warned Meta in April and subsequently deleted their data. By October, the company had implemented stricter rate limits to prevent such mass enumeration.

Meta referred to the exposed information as "available public base data" and stated that it found no evidence of malicious exploitation. The vulnerability was known: in 2017, Dutch researcher Loran Kloeze had already described the same technique in a blog post.

The researchers collected 137 million phone numbers from the U.S. and nearly 750 million from India, as well as 2.3 million numbers from China and 1.6 million from Myanmar, even though WhatsApp is banned in both countries. Analysis of cryptographic keys also showed the existence of duplicate keys, likely generated by unauthorized WhatsApp clients, not due to a platform failure.

Nvidia sees an AI transformation, not a bubble, while skeptics count the risks

Nvidia CEO Jensen Huang argues that we are not witnessing an AI bubble, but rather a historic inflection point. He believes that the type of computing in which Nvidia excels will become ubiquitous, from coding to controlling entire fleets of robots in the real world.

On the other hand, more and more skeptics in the markets fear that from such a "peak," the only possible direction is down. Nvidia recently published results and forecasts that exceeded expectations, calming investors in the short term, but long-term concerns remain.

A regulatory document shows that 61% of the company's $57 billion revenue in the third quarter comes from just four large clients, unnamed. Huang speaks of three major transitions: the migration of non-AI software from traditional processors to Nvidia's high-performance chips, the emergence of new categories of software such as code assistants, and, ultimately, the expansion of AI from the virtual world into the physical one, through machines and robots.

All these plans involve building a huge number of data centers, which will consume massive amounts of land and energy, raising concerns even among the most optimistic supporters of the company.

The European Commission delays full implementation of the AI Act until 2027

The European Commission has presented the "Omnibus Digital," a package aimed at simplifying the EU's major digital regulations. Among them is the AI Act, adopted last year, which was to be implemented gradually.

The new proposal announces a delay for some of the rules applied to the riskiest forms of AI. In particular, companies and organizations using "high-risk" AI systems for analyzing CVs, assessing exams, or approving loans will have more time to adapt. These systems will only fall under the full scope of the AI Act in December 2027, over a year after the initial deadline of August 2026.

The Commission justifies the delay by the lack of implementation at the member state level and the need for companies to have additional time to comply with very complex rules. The tech lobbying association CCIA, which includes companies like Amazon, Apple, Google, and Uber, welcomed the delay but called for even more "courageous" and "clear" measures.

Critics argue, however, that the pause goes too far and undermines the promised protections. While countries like France and Germany have welcomed the AI Act, many experts in AI ethics and safety regret that numerous member states have failed to create the necessary enforcement structures in time for the initial schedule.

Google Gemini adds AI image detection with SynthID

Google is launching in the Gemini application a tool for verifying AI-generated images, allowing users to find out if a visual was created or modified using Google technology. Gemini checks for the presence of invisible watermarks from SynthID and provides more context about the image.

Users can upload a photo in the app and ask if it was created with Google AI. SynthID inserts imperceptible signals into AI-generated content, and Google states that over 20 billion pieces of content have been marked this way since 2023.

Gemini will combine SynthID verification with its own reasoning to provide detailed answers about content encountered online. For images created with the free Nano Banana Pro and Google AI Pro, Google will also keep a visible watermark (the Gemini glow), so it is clear that the image is AI-generated. For Google AI Ultra users, the visible watermark will be removed to provide a clean visual canvas suitable for professional work.

Google plans to extend SynthID verifications to video and audio and to add C2PA content credentials for more images.

AI is too risky to be insured, warn insurance companies

Major insurers like Great American, Chubb, and W. R. Berkley are asking U.S. regulators to allow them to exclude AI-related liabilities from corporate policies. One underwriter described the outcomes of AI models as "too much of a black box" to be assessed in a traditional manner.

The industry has serious reasons for concern. One example: Google AI Overview wrongly accused a solar company of legal issues, leading to a defamation lawsuit. In another case, a Waymo vehicle hit a pedestrian, triggering another lawsuit. Insurers fear that a single AI-related catastrophe could deplete the reserve capital accumulated by companies.

While firms seek policies to cover their AI risks, insurers say the technology is too unpredictable. A study conducted by Everest Group, together with actuarial technology company Akur8, shows that current underwriting methods for AI risks "are not sufficient" to cope with the real complexities.

Major consulting firms WTW and Mercer warn that excluding AI from general liability policies would leave most companies "uninsured" for associated risks. The lack of solid actuarial data and the rapid pace of technological evolution make it extremely difficult to estimate potential losses, forcing insurers to choose between underestimating the risk and overcharging for it.

Synthesis made with the help of a monitoring feed provided by Control F5 Software.