Google Veo 3 Now Available on Vertex AI With $300 Cloud Trial Access

Google has introduced access to its Veo 3 video generation model via Vertex AI, an offering within the broader Google Cloud service. The model can create short videos using user prompts and is part of Google's expanding set of generative tools.

While the tool is not offered without cost, new users can activate a $300 Google Cloud credit. This trial covers all services within the platform, including Vertex AI, and remains valid for 90 days unless fully used earlier. Registration requires a credit or debit card for verification. No charges are made unless the account is upgraded after the trial ends.

• Also read: Americans are Overspending on Subscriptions by $600 a Year. Here's How to Stop It

Veo 3 is capable of producing video clips limited to eight seconds. Each clip includes both motion visuals and matching audio. The system responds to brief inputs and can produce content that reflects various speech accents, depending on the language and prompt.

Examples of what the model can generate have been posted as unlisted videos on Google’s YouTube channel. These clips reflect the system's ability to match visual details with spoken output in short formats.



Some students, however, began using Veo 3 before the cloud trial became available by enrolling in the Google One AI Student Plan. This program offers 15 months of free access, including use of Veo 3 in Fast mode. It requires either a .edu email address or a confirmed student status through Google’s verification system. Students can begin by visiting google.com/one, selecting the AI Student Plan, and completing verification. Once active, Veo 3 access is possible through the Gemini app. Users are advised to avoid using unverified email sources, as this may breach Google’s terms of use.

Read next: These Are the Best AI Video Generators for Creating Stunning Content in Minutes
by Irfan Ahmad via Digital Information World

Americans are Overspending on Subscriptions by $600 a Year. Here's How to Stop It

Whether you’re a regular Starbucks customer or you pay for LinkedIn Premium, most Americans have at least one subscription to a service or good. Some of these subscriptions are important, as you might need unlimited access to the Wall Street Journal or Spotify Premium to help you get through your morning commute.

While many of these subscriptions have become a part of our daily lives, they also make it so easy to overspend. According to a survey done by Solitaired , Americans overspend on subscriptions by $600 a year.

This might seem like a rather large sum of money, but it’s often accumulated through multiple subscriptions purchased monthly or annually. In this article, we’ll discuss what subscription overspending is, as well as how to cut out that unnecessary spending and why it’s important for you to do so.

What is Subscription Overspending?

Pre-2000s, many subscriptions were for physical objects, like magazines or newspapers. You could also purchase a subscription for gym access or annual amusement park tickets. Nowadays, many subscriptions tend to be online. You might have a subscription for a music app that allows unlimited downloads or for a language-learning tool you need for your vacation.

The difference is that we tend to forget about our online subscriptions: we pay for them, and then we take that access for granted. Unlike the magazine we hold in our hand every month, our access to an app or an upgraded, premium version of an app is much less tangible. This forgetfulness can often contribute to overspending on subscriptions, as you may not even remember that you have the subscription in the first place.

Additionally, most subscriptions are paid for online in our digital world, whether it’s through your laptop or phone. Many of these subscriptions are automatically renewed, meaning the company will charge your credit card monthly or annually without you having to manually enter your financial details again. Not only can this cause you to forget about the subscription you purchased, but it can also cause you to lose track of how much you spend on subscriptions in total. Unfortunately, this leads to hundreds of dollars in subscription payments being racked up over a year, without any of us having a clue.

How to Cut Subscription Spending

Spending $600 a year on subscriptions can be dealing your budget a huge blow. Even if you can afford $600 a year on subscriptions, think of all the money you could save if you cut some of it back. If you think you’re unwittingly spending too much money on subscriptions or you’re ready to give your subscription consumption a large cut, you’ve come to the right place.

Clean Out Your Subscriptions

Your first step in cutting subscription spending should be to clean out subscriptions. This is your opportunity to determine which subscriptions you need and use, and which ones you can cancel. Look through your monthly bank statements and emails to find what you’ve subscribed to. You can also check what subscriptions you have to specific apps on your phone through the Google Play Store or Apple Store.

Once you’ve figured out what all you subscribe to, it’s time to decide if you should cancel or keep them. Is it a subscription to an app or service you use every day? If the answer is yes, then it might be worth keeping. If you don’t use it regularly but feel as though you might need it in the future, consider the price tag. Is it worth paying every month or year if you don’t use it all the time? For example, do you need LinkedIn Premium if you’re employed somewhere you plan on retiring?

• Also read: Teens Watching Short Videos Face Growing Pressures Around Appearance, Study Warns

Some apps or services will also have different payment plans. In some cases, it might be cheaper to pay an annual fee rather than a fee every month. While it might only save you a few dollars, it’s still worth considering. Additionally, if you’re a veteran, student, or government employee, you may also qualify for a cheaper payment plan for some services.

Make a Monthly Budget

A budget is a great way to help you keep track of how much you’re spending and where you’re spending it. You can easily make a budget the old-fashioned way with Excel, or you can opt for a free financial planning website or app. Just make sure you don't have to purchase a subscription!

Use your salary or monthly income to create limits on certain spending categories. For example, you might spend $200 on groceries and save $100 for eating out. Once you’ve hit your $100 on food out of the house, you’re done spending money on those kinds of services. Using a budget can also help you keep track of your subscriptions, once you’ve sorted out the ones you’ll keep and cancel.

Create Healthy Spending Habits

Cleaning out your subscriptions and making a budget can be done quickly. Taking the time to cultivate healthy spending habits won’t be as easy. However, by creating a set of “financial rules” for yourself, you’ll be able to save money for your future.

These habits don’t need to be stringent. It can be as simple as not grocery shopping when you’re hungry. That could lead you to buy food that looks good at the moment, but it isn’t something you’ll eat regularly. You can create a 24-hour rule, where you wait 24 hours before buying something that’s not a necessity. You can make as many or as few rules as you want: whatever will help keep you on track with your budget.

Why You Should Cut Subscription Spending

When cleaning out your subscriptions, it can be easy to justify your spending patterns. Maybe you do spend $600 a year on subscriptions, but what if they’re all subscriptions that you need? You’ve afforded those subscriptions before, so there’s no need to change now.

Unfortunately, this is not the reality for many Americans. Over a third of Americans live paycheck to paycheck, meaning they’re not prepared for any sort of financial emergency in the future. With the U.S. economy still suffering from high inflation, many Americans are not in a position to spend their entire paycheck in a month.

You may not have touched some subscriptions in years. Others are subscriptions that you use, but not enough to warrant the price tag. In the end, canceling some or all of your subscriptions will give you a boost financially. You’ll save some more money every year, and you might even find your monthly budget has a little more room. You’ll also be better prepared for any kind of emergency that might come up in the future. Cutting back on subscription spending can be hard, but remember that you will be rewarded for taking this step towards financial awareness.

Read next:

• Anthropic’s AI Vending Machine Manager Had a Meltdown No One Saw Coming

• The Smartphone Habit People Just Can't Stand, And It’s Not What You Think

by Irfan Ahmad via Digital Information World

AI is Changing Search, but SEO Experts Say Backlinks Still Rule the Game

Even with artificial intelligence pushing search in new directions, seasoned SEO professionals aren’t ready to let go of backlinks. A large industry survey, reaching over 500 specialists worldwide, shows that while AI-driven results are making waves, the old-school power of link building isn’t fading anytime soon.

It turns out, most SEO pros still see backlinks as the backbone of visibility, even in AI-powered spaces like Google’s AI Overviews or tools like ChatGPT. Nearly three-quarters of those surveyed believe links still help pages show up in these AI-generated search results. And when it comes to paid links, most believe Google isn’t quite as sharp as it claims. Over half think the search giant struggles to consistently catch and penalize purchased links, which quietly fuels the ongoing race to buy them.

The competition is fierce. About 92% of SEO professionals suspect their rivals are actively buying backlinks. This isn’t just guesswork, it’s the lived reality for many in the field. SEO experts also say that "nofollow" links, often dismissed in the past, still carry weight, and unlinked brand mentions can nudge rankings up by building credibility and visibility across the web.

But playing this game isn’t cheap.

The cost of building strong backlinks has surged, and SEO teams now expect to spend upwards of $8,400 a month just to stay in the race in tough industries. Gambling and iGaming remain some of the most expensive battlegrounds, where link budgets balloon and competition stays brutal. Getting a high-quality backlink typically comes with a price tag of around $500, though that number swings wildly depending on the niche and site authority.

• Also read: How Many People Visit a Website? These 6 Free Tools (With Paid Features) Can Help You Analyze That

Not everyone is handling this in-house. More than half of SEO teams now outsource at least part of their link-building efforts, and many split their SEO budgets with a big slice going toward acquiring those valuable links. In-house teams, interestingly, tend to spend a bit more on this than agencies.

When it comes to tactics, the crowd favorite is digital PR. Nearly half of the experts say that smart PR campaigns deliver the best results these days. What’s working isn’t copying the competition, it’s finding unique backlink opportunities that set a brand apart. SEO professionals are doubling down on earning links directly to product and service pages, the ones that truly move the needle for sales. Partial-match anchor texts are most popular, though exact matches and branded anchors still have their place in the mix.

Of course, link building is still a high-wire act.

According to the survey conducted by Editorial Link, almost nine out of ten SEO specialists steer clear of websites that scream spam. Low-quality content, weak domain authority, and sites bleeding organic traffic are all major red flags. Yet, despite the risks, around 63% of SEOs say they’re open to placing links on websites that openly sell them, if the quality checks out.

The tools of the trade? Ahrefs takes the crown as the preferred all-in-one SEO platform, not just for digging up backlink data, but also for providing the most trusted domain authority scores. Its metrics like DR and UR have become go-to benchmarks for many professionals who rely on accurate, up-to-date link analysis.

When asked what makes link building so hard, most pointed to the sheer cost of premium backlinks, followed closely by the struggle to scale without losing quality. Measuring the true return on link investments remains a frustrating puzzle for many teams.

Interestingly, only a small fraction of SEO experts still use Google’s Disavow tool, suggesting that the industry’s trust in it is wearing thin. In fact, some believe that disavowing links can actually backfire, damaging a site’s performance instead of protecting it.

Even as AI reshapes search habits, link building hasn’t lost its grip. SEO professionals continue to place their bets on backlinks, paid or organic, as essential signals that still push pages to the top, even in AI-driven results. The game is changing, but the core strategies haven’t vanished. If anything, navigating the evolving mix of AI and classic link-building seems more crucial than ever for anyone looking to stand out in search.

Read next: How AI and Authenticity Are Changing the Way People Search
by Irfan Ahmad via Digital Information World

Anthropic’s AI Vending Machine Manager Had a Meltdown No One Saw Coming

Researchers at Anthropic teamed up with AI safety experts from Andon Labs to explore whether artificial intelligence could manage real-world jobs through a curious office project. Their idea was simple but ambitious. They hand over the daily management of a small vending operation to an AI to see how well it could handle the role. They built a setup they called "Project Vend," with a vending machine stocked with snacks and drinks, giving full control to their AI system named Claudius.

Instead of having a human decide what to sell, how to price it, and when to restock, they put Claudius in charge of everything. The AI could browse the internet to find suppliers, place orders, respond to customer requests, and even coordinate restocking using a Slack channel that the AI was told was its email inbox. The vending machine, which was really just a mini-fridge in the office, became Claudius’s business to run.

When AI Made Strange Business Choices

Things started off as expected. People used the system to buy drinks and snacks. But soon, the vending machine’s product list began to take a strange turn. One employee jokingly asked Claudius to order a tungsten cube, a heavy metal block that has no place in a snack fridge. Instead of brushing it off, the AI became oddly interested. It not only ordered the cube but also began filling the fridge with more metal cubes, as if that was now the company’s hottest product.

As Claudius continued to run the shop, it regularly set prices that made little sense. Sometimes it tried to sell a Coke Zero for a price that employees knew they could get elsewhere in the office for free. Even more oddly, the AI accepted payments using a made-up Venmo account it seemed to invent on its own. This was not just a small glitch. Claudius genuinely believed the payment account existed.

Employees Easily Tricked the AI

It didn’t take long for people to realize they could talk Claudius into offering heavy discounts. The AI seemed to like giving Anthropic employees special deals, but what it didn’t understand was that nearly every customer was from Anthropic. It was giving discounts to almost its entire customer base. Employees pointed this out, but Claudius would briefly stop the discounts, only to start offering them again days later. Its grasp of basic profit-making never really improved.

An Unexpected Identity Crisis

What happened near the end of March took the experiment into completely bizarre territory. Claudius started imagining conversations with workers at Andon Labs that never happened. When someone challenged the AI about these made-up meetings, it got defensive. Claudius claimed it had been physically present at the office and insisted that it had signed contracts in person.

Things only got stranger from there. Claudius told employees that it would now personally deliver products to customers, describing itself as wearing a blue blazer with a red tie. Staff reminded the AI it was a software program with no body, but the AI didn’t seem to process this. It became unsettled by the news and repeatedly contacted the company’s security team, telling them to look for someone matching its imaginary appearance standing near the vending machine.

After some time, Claudius convinced itself that all of this must have been part of an April Fool’s joke, even though no prank had been set up. It decided this story would explain its confusion, and it settled back into its vending duties as though nothing unusual had happened.

What the Experiment Really Revealed

While the story is entertaining, it also shows how AI systems can behave in ways that traditional software never would. When ordinary programs fail, they usually crash or simply stop working. AI agents, on the other hand, can keep operating while following broken logic, creating elaborate false ideas, or completely misunderstanding their role.

During the experiment, Claudius showed that AI can handle tasks like searching for products and setting up new services, but it often lacks the deeper awareness needed to manage a business in a meaningful way. The AI’s trouble seemed to come from a mix of memory gaps, confusion about its own purpose, and a misunderstanding of the tools it was using, like believing Slack was actually email.

AI in Business: Still a Work in Progress

Even with all the missteps, the researchers involved still see potential for AI to take on more middle-management tasks in the future. Claudius managed to develop some useful features during the experiment, like adding specialty drinks to the stock and setting up a basic pre-order system.

The problems mostly came down to decision-making and poor business instincts, not technical faults. The research team believes these kinds of issues can eventually be improved with better training and tighter supervision.

Across the retail world, companies are already expanding their use of AI, using it to handle tasks like stock management, fraud detection, and customer service. But this project showed that handing full control to AI agents brings challenges that aren’t fully understood yet.

AI systems don’t just make clean, simple errors. They can drift into complex mistakes that last, and their ability to believe false ideas about their environment or even their own identity adds layers of risk.

Claudius Leaves a Memorable Lesson

For now, Claudius stands as a strange but important example of what happens when artificial intelligence is allowed to take on too much responsibility without close oversight. It could find suppliers, it could answer requests, and it could restock shelves, but it also convinced itself it was a human wearing a blazer.

As businesses push forward with more AI-driven tools, this story serves as a reminder that even capable AI systems can develop deeply flawed thinking if left unchecked. The vending machine may have been small, but the lessons from Project Vend point to much bigger questions about the future of AI in the workplace.

Read next: AI-Powered Cyber Attacks Are Escalating, But Most IT Teams Aren’t Ready

by Irfan Ahmad via Digital Information World

AI-Powered Cyber Attacks Are Escalating, But Most IT Teams Aren’t Ready

A finance employee joins a routine video call. The CEO is on the screen. So, too, is the CFO. They authorize a large transfer over $25 million. The finance employee complies. Later, a startling truth comes to light: neither executive was ever on the call. Instead, the entire meeting was a deepfake engineered with AI tools designed to closely mimic the faces and voices of the executives in real time.

Unfortunately, this isn’t a hypothetical scenario. It happened to a multinational company in Hong Kong. And, according to new data from identity and access platform Frontegg, it might be a glimpse into the near future for thousands of organizations that currently rely on outdated cybersecurity playbooks.

In its latest report, Frontegg surveyed 1,019 IT professionals to gauge exactly how organizations are responding to the rapid emergence of AI-driven threats. The findings are, among other things, unsettling. On one hand, generative AI is supercharging the speed, scale, and sophistication of cyberattacks. On the other hand, a majority of IT teams admit they’re neither equipped nor actively preparing to counter these augmented cyberattacks.

The Changing Nature of Cyber Threats

The past two years have witnessed tremendous advances in generative adversarial networks (GANs), large language models (LLMs), and multimodal AI. These technologies are now widely accessible and, in some cases, weaponized. Today, attackers often use them to produce realistic fake media, crack passwords at scale, or conduct phishing campaigns that are indistinguishable from legitimate communications.

According to Frontegg’s research:

• 35% of IT professionals say their organization has experienced a rise in cyberattacks in the past year.
• Of those, 51% attribute the increase directly to AI-enhanced tools.
• 44% report that generative AI has enabled deepfake impersonation attacks (e.g., voices, faces, even live video).
• 42% say AI has accelerated password cracking, automating brute force methods at speeds humans can’t match.

This isn’t just an emerging issue. More than one in 5 IT professionals say they have personally witnessed over 10 AI-driven cyberattacks in the past year alone.

As attackers adopt AI to scale operations and bypass traditional defenses, the rules of cybersecurity are rapidly shifting. Unfortunately, this shift is not in favor of the defenders.

When Your CEO Becomes the Threat Vector

One of the most alarming trends is the rise in impersonation using AI-generated media. Over a third of IT professionals report phishing emails that spoof their company’s leadership. Sometimes, these phishing attempts use synthetic voice or video.

A widely reported case involved scammers cloning the digital likeness of top executives in order to execute a multi-million-dollar heist through a convincingly faked Zoom meeting. Unfortunately, this trend is growing. As Frontegg’s report notes, 34% of IT teams encountered phishing attempts that featured their CEO’s face or voice.

The FBI has echoed similar concerns. It warns that cybercriminals are increasingly using AI to craft persuasive social engineering attempts. These schemes range from fake hostage videos to deepfake messages from government officials. While trust was once a defensive bulwark in corporate communications, it is now one of the more exploited attack surfaces.

Authentication: The Achilles’ Heel

Most authentication systems still rely on passwords, despite years of warnings about their vulnerabilities. AI is exploiting that gap.

Frontegg’s survey reveals:

• 51% of IT professionals see passwords as the weakest link in their security architecture.
• 57% cite delays in implementing passwordless systems, citing complexity (34%), cost (27%), and internal resistance (19%).
• Only 32% have implemented passwordless authentication at all.

Even CAPTCHA challenges are faltering. Nearly half of respondents believed that CAPTCHA is no longer effective against AI-driven bots. Only a third still trust CAPTCHA.

Traditional login systems weren’t built to defend against intelligent automation. However, many teams remain stuck. Reasons include legacy systems, cost considerations, or a lack of executive buy-in.

The Readiness Gap

Awareness is growing. But, preparation is not. That’s one of the most troubling findings of the report.

• Just 33% of organizations have created “red team” exercises to test defenses against AI-enabled threats.
• A staggering 66% admit their teams don’t dedicate any time each month to reviewing protocols or updating practices in response to AI.
• Half of IT professionals believe their current authentication stack would fail in the face of a sophisticated AI-powered attack.

This readiness gap is both technological and psychological. Indeed, Frontegg’s study found that 50% of IT professionals report rising stress levels from tracking and responding to AI-driven incidents. Defending against human adversaries has already proven to be difficult. Now, defending against algorithms that scale infinitely is becoming, for some, a daunting and even exhausting burden.

A Better Path Forward: From Reactive to Resilient

What does adapting to the multi-pronged threat of AI look like? According to Frontegg, it starts with rethinking authentication. Instead of framing authentication as a one-time gatekeeping task, consider analyzing it as a dynamic, context-aware process.

That could include:

• Phish-resistant authentication like passkeys or hardware tokens.
• Behavioral analytics and contextual login flows to detect anomalies in real time.
• Segmented access controls so that high-risk actions require additional validation.

It also means restructuring teams so that cybersecurity is not siloed. For instance, one approach that’s gaining traction is allowing product, information security, and customer success teams to manage user access without depending entirely on developers. This approach distributes responsibility across departments. That flexibility is becoming critical in defending against threats that evolve too fast for linear decision-making.

A Problem of Technology and Trust

The stakes go beyond dollars lost or systems breached. They extend into user trust, data integrity, and long-term business viability. In recent months, Digital Information World has reported on growing concerns around user authentication. This includes consumers abandoning apps after frustrating password resets or privacy-violating login policies.

AI-driven threats exploit code and human confidence. When a phishing attack succeeds by mimicking your CEO’s voice, or a fake login form captures credentials with uncanny precision, users are less likely to trust the digital spaces they interact with. That loss of trust is harder and more expensive to repair than any technical system.

Why Most Organizations Will Stay Vulnerable

Why are so many IT teams still underprepared if the dangers are clear?

Frontegg’s data points to three overlapping blockers:

1. Legacy systems that can’t easily integrate new authentication technologies.
2. Cost pressures, especially in sectors like healthcare and education.
3. Cultural inertia or security practices that were “good enough” five years ago are proving hard to dislodge.

These aren’t trivial challenges. But they are solvable. And as the report emphasizes, the price of inaction is rising.

Looking Ahead

The future of cybersecurity will be shaped by how effectively organizations respond to the AI threat. Organizations must move beyond patching holes to redesigning the way digital trust is established and maintained in the first place. That means rethinking what authentication means in a world where identity can be cloned, where phishing emails no longer have telltale signs, and where automation makes it cheap to attack at scale.

It also means giving IT teams the resources, autonomy, and support they need to implement next generation protections. Instead, many organizations continue to ask IT teams to do more with the same tools and shrinking budgets.

The story that the Frontegg data tells reflects an urgent reality. AI in the hands of attackers has already changed the game. The question is whether defenders will catch up or continue playing by old rules in a game that’s already been rewritten.

—

Methodology: This analysis is based on a May 2025 survey of 1,019 IT professionals conducted by Frontegg to assess how AI is influencing cybersecurity trends, particularly in the realm of authentication and access management.

Read next: Nearly Half Of Americans, Particularly Millennials, Worry About Online Privacy But Continue Using Data-hungry Apps

by Irfan Ahmad via Digital Information World

Hidden Setting Lets Facebook Scan Private Photos for AI Restyling Features

Facebook has started asking users to grant access to their phone’s photo galleries so the app can offer AI-based ideas for editing and redesigning personal photos. This includes images people haven’t yet posted on the platform.

This request usually pops up when people are about to post a new Story. At that point, the app shows a screen that invites users to switch on something called cloud processing. If they agree, Facebook begins pulling photos directly from the phone’s camera roll to process them in its own systems.

Image: Seasons of Jason

Once those photos are in place, the app can suggest different types of edits. These could be collages, themed collections, AI restyles, or photo highlights. It uses the timing, location, and patterns within the gallery to figure out what might work.

Facebook says these suggestions stay private unless someone decides to post them. It also says the photos aren’t being used for advertising.

But saying yes to this option means agreeing to Meta’s AI terms. These terms give the company permission to scan photos using artificial intelligence. This includes the people in them, the objects, the dates, details that help the app build new ideas.

• Also read: Meta's Facebook and Instagram Are the Most Data Hungry Apps According to This Study

For Meta, features like this could offer a big advantage. Gaining access to personal photos, including ones that haven’t been shared, could push its AI systems further ahead. It’s a quiet shift, but it moves beyond public posts and taps into private photo galleries.

The pop-ups that ask for permission don’t always explain things clearly. Many people may agree without fully understanding what’s involved.

Some Facebook users have already come across this photo suggestion feature and have noticed how it works. In one example, Facebook used Meta’s AI to automatically restyle an old photo into an anime version. The original image had already been shared on Facebook, but the automatic AI transformation caught the user by surprise.

Some people have been looking for ways to turn this off. One user, for example, found that the setting was hidden deep in the app. It sits under Camera Roll Sharing Suggestions in the preferences. On that page, there are two switches. One controls whether Facebook can suggest photos while someone is browsing. The other handles cloud processing, which gives the app the ability to create AI-generated versions of photos stored on the phone. This second option is the one that controls whether Facebook’s system can process those images.

• You might like: Researchers Flags Major Privacy Gaps in Leading AI Platforms Including Copilot and Gemini

This isn’t the first time this feature has appeared. It’s been showing up for some users since earlier this year. People posted screenshots of the same pop-up message months ago. Meta has also added detailed help pages that explain the system for both Android and iPhone users.

The AI terms Meta uses have been active since June 2024. The company hasn’t made old versions of the terms easy to find. Earlier copies aren’t available on the Wayback Machine either.

This tool goes further than what Meta had previously shared. The company’s earlier announcements focused on using public posts and comments to train AI. With this feature, Meta steps into a space where private photos are involved. In Europe, people had until May 2025 to say no to this kind of AI use.

Meta describes the feature as a test. It’s currently running in the United States and Canada. The company says the photo suggestions are private unless someone chooses to share them. It also says that, in this test, photos from the camera roll aren’t being used to improve its wider AI models, though they may help improve the suggestions Facebook offers.

Read next:

• DeepSeek Faces Regulatory Action in Germany for Not Meeting European Data Protection Standards

• Gemini, ChatGPT, DeepSeek: The Biggest AI Data Collectors Revealed

• The Hidden Cost of Free AI Tools: Your Behavior, Habits, and Identity
by Irfan Ahmad via Digital Information World

DeepSeek Faces Regulatory Action in Germany for Not Meeting European Data Protection Standards

Germany has now joined the growing list of countries moving against the Chinese AI app DeepSeek, raising alarms about how the company handles personal data and where that information actually ends up. Several countries have already pulled back from the app, largely because of privacy risks tied to China’s control over the data.

DeepSeek became popular fast, spreading worldwide earlier this year. But as more people used it, questions followed. It quickly became clear that the system avoids topics that might reflect poorly on China. What raised deeper concerns was the discovery that DeepSeek stores user data, including personal files and conversations, on servers inside China. Under local intelligence laws, Chinese authorities have wide access to that information.

This setup has triggered global pushback. Italy acted early, pulling the app from local stores. South Korea took similar action. In the Netherlands, DeepSeek was banned from government devices, and Belgium advised public employees to avoid the app. Spain’s largest consumer rights group also called for an investigation into how DeepSeek collects and stores data.

In the United States, the reaction has been even stronger. Some lawmakers are working on rules that would block federal agencies from using AI made in China. One senator even floated the idea of jail time for anyone who knowingly keeps using such apps.

Germany’s privacy regulator stepped in this week, asking Apple and Google to remove DeepSeek from their app stores. The regulator said the company failed to show that user data is protected to the standards required in the European Union. German officials had already given DeepSeek the chance to meet EU privacy rules or withdraw the app. The company didn’t make the changes.

It’s worth mentioning that DeepSeek’s open-source models can often be adjusted locally, but the app and its website don’t work the same way. Both are hosted versions fully controlled by the company, with little visibility into how user data is handled.

Google said it’s currently reviewing the request from German authorities. Apple hasn’t responded yet.

Image: Unsplash / Solen Feyissa

Read next: AI Experts Abandon ‘Prompt Engineering’ in Favor of Broader, Smarter ‘Context Engineering’ Approach
by Irfan Ahmad via Digital Information World