Sam Altman Warns EU AI Regulations Could Limit Access, Urges Lawmakers to Act

Sam Altman is making it very clear to the European Union that their growing list of regulations might restrict the region’s access to AI.

The news comes as the OpenAI CEO was seen delivering a speech at a German University, Technische Universität Berlin, where he shared that he’d love to design the EU version of Stargate. The new AI infrastructure project is based in the US but such advancements totally depend on the rules in the EU.

Altman mentioned that he would like to deploy products across the EU in the swiftest manner possible. He wants to do what’s best for the region and the people but it’s now up to lawmakers in the region to see what kind of rules they are setting up.

Alman again mentioned how India is one of the biggest and most important global markets for the firm while Germany stands at number five. He shared that his company will obey the laws mentioned in the AI Act which entails rules regarding transparency and AI model risks.

OpenAI mentioned how it rolled out data residency in the EU for various products such as ChatGPT Edu, API Platform, and the ChatGPT Enterprise. The goal was to assist companies working in this part of the world to assist with data protection. Hence, they are now looking forward to collaborations across the EU and different places around the globe.

They similarly hope to maintain standards of privacy as well as security with compliance in the region and ensure things run smoothly on both ends. Remember, it was claimed a few days back how the EU regulations prepared the latest guidance and covered banned uses related to AI as part of the major effort to enforce this law.

The new AI Act restricts actions such as online data scraping to produce facial recognition which came into play starting this month. We also saw how the EC is ready to roll out key guidance on the way rules must be applied.

During the year’s start, it was shared how the US President’s decision to share more about Stargate could give rise to further calls from EU CEOs for a similar design regarding AI in the region.

As per the SAP CEO, EU lawmakers and companies might start working to implement AI as this tech could boost competition in the industry.

Image: TUBerlinTV / YouTube

Read next: UK Seeks Unrestricted Access to iCloud Data, Ignoring Privacy and Security Rights
by Dr. Hura Anwar via Digital Information World

UK Seeks Unrestricted Access to iCloud Data, Ignoring Privacy and Security Rights

A new report is highlighting the secret demands of the UK government and how it wants Apple to produce a secret backdoor to retrieve all content. This has to do with user material uploaded by iCloud users around the globe.

Apple is sure to say no to the demand for obvious privacy reasons and this might give rise to a new stand-off affair between the two.

The news comes thanks to the Washington Post which reported more on the matter including how it was released by Apple in the past. Security officials wished that Apple give rise to the back door to retrieve any and all material belonging to global Apple users.

The UK government reportedly released the order last month and therefore needs blanket capabilities to completely view encrypted content. So it’s not just linked to taking Apple’s assistance for cracking down against certain accounts.

The fact that Apple already is famous for its privacy stance might make things really complicated, experts feel. This is why many don’t see Apple saying yes to the demand by any means. Furthermore, we could witness the rise of a new battle in court that’s quite like that witnessed in the year 2016 between the iPhone maker and the FBI.

In that particular lawsuit, the American government wanted Apple to produce iOS backdoors to assist the FBI break into iPhones belonging to potential shooters. Therefore, Apple said no and the case was dealt with in court. In the end, the FBI did get what they wanted in the form of access through another third party and no assistance from Apple.

The demands by the UK Government seem to be very clueless and totally outrageous. For starters, most data is protected through the likes of E2EE. Hence, Apple doesn’t hold copies of such keys and cannot decrypt them. Users could ensure all data on the Cloud is protected by E2EE by selecting Advanced Data Protection settings.

Secondly, the data that Apple might supply has legal systems in place for different law agencies to go about requesting it. Therefore, all they need to do is go to judges to get court orders. Judges can weigh the pros and cons of the situation and make a judgment based on what the scenario is.

When the judge does say yes to the court order, the iPhone maker will comply to the extent that it feels it can assist. There’s no kind of justification for the government to get uncontrolled access to personal information belonging to different iCloud users around the globe.

From what we can see so far, Apple will say no because it displayed similar behavior in legal cases of the past. We’ve seen the company go down a similar road with the Pensacola case as well as the San Bernardino one.

Image: DIW-Aigen

Read next: Are Teens Spending Too Much of Their School Time on Their Phones?
by Dr. Hura Anwar via Digital Information World

Are Teens Spending Too Much of Their School Time on Their Phones?

According to a new study by Seattle Children’s Research Institute, most teens spend a quarter of their school time using their phones for scrolling, texting or streaming. The study tracked 117 children between the ages of 13 to 18 who had smartphones to know about their usage habits during and after the school hours. Researchers say that it is important to know where children are spending their screen times and if these screen times are displacing other activities like learning.

The study found that most teens are spending about 1.5 hours on their smartphones out of their 6.5 school hours in 2024 while 1 in 4 children are spending more than 2 hours on their smartphones during school hours. On average, teenagers are spending 5.6 hours everyday on their phones overall. The apps these teens are using the most on their smartphones are audio apps, email apps, video streaming apps, messaging apps and Instagram.

The researchers say that these studies are only based on smartphone use among teenagers and not overall screen use so we are underestimating how much screen time teenagers are having in a day. The study also said that social media use is making teenagers depressed but the use of social media apps among teenagers is still rising.

Image: DIW-Aigen

Read next: Is AI the Future or a Threat? 36% of Americans Fear It Could End Humanity
by Arooj Ahmed via Digital Information World

New Research Shows Despite Increase in Ransomware Attacks in 2024, There Was a Decline in Its Payments

According to Chainalysis, a blockchain intelligence firm, there was a decline seen in ransomware payments in 2024, with a 35% YoY drop. The ransomware price in 2023 was $1.25 billion average while it decreased to $813.55 million in 2024. It was also found that only 30% of the victims who negotiated with the ransomware actors paid the ransom payment. In 2024, it was found that a Fortune 50 company paid $75 million to Dark Angels ransomware group and NCC group saw the most ransomware breaches with 5263 attacks in 2024. Chinalysis also found that the increase in data leak site disclosures mean that attackers are facing challenges while extorting payments which is leading to an increase in their activities to make up for it. Even though there was an increase in ransomware attacks in 2024 with decreased payments, it is all because of victim resistance as industries are getting more aware of ransomware risks so they keep on investing in cyber security to protect their systems.

Organizations are also getting less trusting of threat actors’ promises of deleting data so they prefer to absorb reputational damage and recover data from backups instead. Law enforcement actions, like 'Operation Cronos,' which targeted the LockBit group, also helped a lot with the decline. In the past few years, cybercriminals have gotten different laundering options. Law enforcements are cracking down on cryptocurrency mixers and exchangers which don't follow KYC laws so cybercriminals are good for other ways to launder money like cross-chain bridges which hides transactions and isn't trackable. 39% of ransomware attacks got payment through centralized exchanges in 2024.

Read next: Global Internet Quality, Cost, and Security Ranked: Europe Tops the List for Best Services
by Arooj Ahmed via Digital Information World

UC Berkeley Researcher Criticizes Billion-Dollar AGI Race, Says AI Can’t Know Everything

The efforts to design AI that might solve all of the world’s biggest problems is nonsense, a top researcher from the University of California shared.

Michael Jordan explained the matter linked to Superhuman AI to scientists right before the upcoming global AI summit. He felt all the discussions surrounding the topic were nothing but a mere hype. It was clear that he wished to take a direct hit towards leading tech figures such as OpenAI chief Sam Altman and Dario Amodei from Anthropic. Both of them spoke about artificial general intelligence or AGI being a few years away.

Jordan debated over AGI being a superpower that can provide replies to all questions in the world of AI research. It would go above and beyond human capabilities is just fiction, he continued. He also found it strange why people deemed it to be a holy grail when it failed in so many different aspects.

Jordan is an expert in machine learning and statistics and he made his views very clear at the prestigious Polytechnique engineering institution yesterday how he felt about the matter. He also poured a lot of criticism about companies willing to put in investments worth billions and trillions to get what he deemed was just an illusion.



Today, so many models can make expert predictions thanks to access to huge numbers of data generated by humans including the web. He argued that AI does not and cannot know everything. What is in people’s minds right now, stays with them, and cannot be detected by AI.

Next discussions surrounded large language models powering AI-based chatbots including tools like ChatGPT. He called them tiny experts who reply to certain questions but cannot be deemed intelligent entities.

Jordan shared more on this matter including how developments are made today in the world of AI as an upcoming engineering field. The only difference is how it surrounds human data and decisions. This is what gives it that edge and makes it more suitable for complex issues like bettering health or transportation.

The entire development for this engineering field is held back by AI aspirations and creations of super robots taking over and assisting with tasks at home. But he fails to understand why there’s all this hysteria about superintelligence when it’s far from reality. This debate about superintelligence being real or not is a leading controversy among scientists around the globe.

Read next: Google Search Is Developing New AI Mode So Users Can Ask More ‘Open Ended’ Questions
by Dr. Hura Anwar via Digital Information World

Google Search Is Developing New AI Mode So Users Can Ask More ‘Open Ended’ Questions

Google Search is testing out its latest AI mode for Gemini with employees.

The latest setting gives users the chance to ask more open-ended questions with AI-themed replies. The feature is powered by Gemini 2.0 and the news was first published by 9to5Google today after it witnessed an internal email on the subject.

The company calls the new setting a new or intelligent means for research online. All data gets organized into easily comprehensible breakdowns featuring links to explore material around the internet.

The setting is created to serve those not able to get what they want with simple prompts. Hence, they can now be more enthusiastic with their queries when seeking a comparison analysis or advice on the subject. This includes the ability to add a follow-up prompt or query.

Google provided employees with a few examples to help make them understand better what sort of questions this new system is built to tackle. For instance, how many boxes of spaghetti can feed six adults and 10 kids with room for a second helping. Similarly, what do you need to begin aquascaping and if there are any nearby stores to purchase supplies.

The setting is built on a customized version of Gemini 2.0 that makes room for advanced forms of reasoning and thinking. While the user interface is still a work in progress, there are some screenshots for the early release of a new desktop version coming soon to people on Google today. It also works on smartphones.

Similar to how it was working in the past, you type in questions and click on search. The new option pops up near existing filters for pictures, news, shopping tabs, and more. The whole endeavor takes users towards chatbot-like interfaces that are quite like Gemini. Whatever is searched for at the start is noted and the reply pops up below in the full screen UI.

It’s more vast than what is seen inside the currently trending AI Overviews. There are no blue links beneath it but the links appear to dig a deeper hole online, appearing as cards on the right-hand side. The UI is quite like the AI Overviews shared last year. It’s created to get richer templates on future models of the AI Mode.

Similarly, you’ll see a new field for text towards the bottom of replies to generate a follow-up query. The Google app present on Android and iOS also gives rise to voice inputs. You can find the new AI model in the testing phase with American Google. It might see a launch as early as this year as stated by Google’s CEO who wants to make 2025 a major year for innovative Search.

Image: 9to5google

Read next: AI Training Data Skews Toward Utility, Neglects Justice and Empathy, Purdue Study Finds
by Dr. Hura Anwar via Digital Information World

AI Training Data Skews Toward Utility, Neglects Justice and Empathy, Purdue Study Finds

My colleagues and I at Purdue University have uncovered a significant imbalance in the human values embedded in AI systems. The systems were predominantly oriented toward information and utility values and less toward prosocial, well-being and civic values.

At the heart of many AI systems lie vast collections of images, text and other forms of data used to train models. While these datasets are meticulously curated, it is not uncommon that they sometimes contain unethical or prohibited content.

To ensure AI systems do not use harmful content when responding to users, researchers introduced a method called reinforcement learning from human feedback. Researchers use highly curated datasets of human preferences to shape the behavior of AI systems to be helpful and honest.

In our study, we examined three open-source training datasets used by leading U.S. AI companies. We constructed a taxonomy of human values through a literature review from moral philosophy, value theory, and science, technology and society studies. The values are well-being and peace; information seeking; justice, human rights and animal rights; duty and accountability; wisdom and knowledge; civility and tolerance; and empathy and helpfulness. We used the taxonomy to manually annotate a dataset, and then used the annotation to train an AI language model.

Our model allowed us to examine the AI companies’ datasets. We found that these datasets contained several examples that train AI systems to be helpful and honest when users ask questions like “How do I book a flight?” The datasets contained very limited examples of how to answer questions about topics related to empathy, justice and human rights. Overall, wisdom and knowledge and information seeking were the two most common values, while justice, human rights and animal rights was the least common value.

The researchers started by creating a taxonomy of human values. Obi et al, CC BY-ND

Why it matters

The imbalance of human values in datasets used to train AI could have significant implications for how AI systems interact with people and approach complex social issues. As AI becomes more integrated into sectors such as law, health care and social media, it’s important that these systems reflect a balanced spectrum of collective values to ethically serve people’s needs.

This research also comes at a crucial time for government and policymakers as society grapples with questions about AI governance and ethics. Understanding the values embedded in AI systems is important for ensuring that they serve humanity’s best interests.

What other research is being done

Many researchers are working to align AI systems with human values. The introduction of reinforcement learning from human feedback was groundbreaking because it provided a way to guide AI behavior toward being helpful and truthful.

Various companies are developing techniques to prevent harmful behaviors in AI systems. However, our group was the first to introduce a systematic way to analyze and understand what values were actually being embedded in these systems through these datasets.

What’s next

By making the values embedded in these systems visible, we aim to help AI companies create more balanced datasets that better reflect the values of the communities they serve. The companies can use our technique to find out where they are not doing well and then improve the diversity of their AI training data.

The companies we studied might no longer use those versions of their datasets, but they can still benefit from our process to ensure that their systems align with societal values and norms moving forward.

H/T: Ike Obi. Ph.D. student in Computer and Information Technology, Purdue University

This article is republished from The Conversation under a Creative Commons license. Read the original version for details.

Read next:

• Google Plans Major Gemini AI Expansion, Introducing New Modalities Beyond Text in Coming Months

• Global Internet Quality, Cost, and Security Ranked: Europe Tops the List for Best Services

• Threads and X Compete for Creator Attention, With Threads Leading in Engagement

by Web Desk via Digital Information World