Kategorie

The Texas Department of Transportation (TxDOT) is warning that it suffered a data breach after a threat actor downloaded 300,000 crash records from its database. [...]

In a twist on typical hiring-related social engineering attacks, the FIN6 hacking group impersonates job seekers to target recruiters, using convincing resumes and phishing sites to deliver malware. [...]

Ivanti has released security updates to fix three high-severity hardcoded key vulnerabilities in the company's Workspace Control (IWC) solution. [...]

Heroku is suffering a widespread outage that has lasted over six hours, preventing developers from logging into the platform and breaking website functionality. [...]

Cybersecurity researchers have shed light on a previously undocumented Rust-based information stealer called Myth Stealer that's being propagated via fraudulent gaming websites. "Upon execution, the malware displays a fake window to appear legitimate while simultaneously decrypting and executing malicious code in the background," Trellix security researchers Niranjan Hegde, Vasantha Lakshmanan Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]

AI acts like Pac-Man—devouring sensitive data across clouds, apps, and copilots. Varonis analyzed 1,000 orgs and found 99% have exposed data AI can access, exposing them to data risks. [...]

OpenAI is working to fix an ongoing outage impacting ChatGPT users worldwide and preventing them from accessing the chatbot on the web or via mobile and desktop apps. [...]

Chinese social media platform RedNote has released its first open-source large language model, dubbed “dots.llm1,” joining a growing wave of Chinese technology companies pursuing open-source AI strategies that challenge Western proprietary models.

The model, developed by RedNote’s internal Humane Intelligence Lab, activates 14 billion parameters out of a total of 142 billion when responding to queries — an architecture designed to balance performance with cost-efficiency.

The company achieved performance comparable to Alibaba’s Qwen2.5-72B after pretraining on 11.2 trillion high-quality tokens without synthetic data, according to the model’s description on Hugging Face.

RedNote, known domestically as Xiaohongshu, operates one of China’s most influential social commerce platforms with over 300 million monthly active users.

A strategic divergence, not just technology

While OpenAI and Google keep their best AI models locked away, Chinese tech companies are taking the opposite approach — giving away their technology for free. But this isn’t just about different business models, according to industry experts.

“Chinese firms like RedNote are deploying open-source LLMs not just as models but as instruments of ecosystem control and geopolitical leverage. Meanwhile, Western firms such as OpenAI and Google remain committed to proprietary architectures,” said Sanchit Vir Gogia, chief analyst and CEO at Greyhound Research. “This is no longer a tactical split in model licensing — it’s a structural divergence in trust frameworks, one that will define the next generation of enterprise AI procurement.”

The split goes deeper than technology choices. “Western AI leaders are optimizing for shareholder return, compliance insulation, and platform lock-in through closed API-delivered models. In contrast, Chinese vendors like RedNote and DeepSeek are aggressively open-sourcing to expand national influence, cultivate developer mindshare, and drive localization-led adoption,” Gogia explained.

Performance vs. purpose

So, how does RedNote’s AI stack up? According to another detailed document on Github, dots.llm1 scored 56.7 on C-SimpleQA, a test of Chinese language skills — not quite as high as DeepSeek-V3’s 68.9, but respectable for a newcomer.

However, some analysts question whether RedNote is playing to its strengths. “Rednote, which has pioneered content-driven commerce, is well-positioned to build a large model based on the ton of data its ecosystem generates,” said Neil Shah, VP for research and partner at Counterpoint Research. “However, in this race of LLMs, Rednote would be better off building a more targeted model to drive AI-driven commerce sitting on a gold mine of data around users’ likes, dislikes, purchase behaviours, etc.”

The economics of free AI

The model uses a “mixture of experts” design that activates only the parts it needs for each task, making it cheaper to run than traditional models. But the economics go beyond technical efficiency.

“RedNote’s dots.LLM1 is less a revenue product and more a market accelerant,” Gogia explained. “The open-source model isn’t a broken business plan — it’s a strategic play to become foundational infrastructure across sovereign cloud ecosystems and public developer communities. The long game is not monetization through licensing but platform entrenchment through adoption.”

This approach is enabled by structural advantages. “Chinese firms benefit from central government subsidies, national procurement incentives, and policy exemptions that support loss-leader behavior in the short term. What might be financially unsustainable in the West becomes strategically viable in China due to alignment with state AI priorities,” Gogia said.

For Chinese vendors, open-source LLMs serve as soft power tools, exporting not only code but also embedded ideologies and governance frameworks into markets across Southeast Asia, Africa, and Latin America.

From social media to a global AI player

RedNote’s AI ambitions go beyond just releasing models. The company has already put AI to work on its main platform with Diandian, a search tool that helps users find content among the platform’s vast trove of lifestyle posts. With nearly 600 million daily searches—about half of what Google rival Baidu handles—RedNote has a real testing ground for its AI.

The timing isn’t coincidental. RedNote grabbed international headlines earlier this year when more than three million Americans joined the platform during TikTok uncertainty, giving the company a taste of global reach. Last week, it opened its first office outside mainland China in Hong Kong.

The trust challenge

For enterprises considering these open-source alternatives, the calculation isn’t straightforward.

“The core trade-off for enterprise buyers is no longer just cost versus performance — it is transparency versus control,” Gogia noted. “Controlled open LLMs provide a bridge across this chasm, giving CIOs and CISOs the ability to audit, customize, and self-host AI models without being beholden to proprietary ecosystems. However, the burden of governance shifts inward — enterprises must build their own trust scaffolding to make openness production-grade.”

The geopolitical dimension adds another layer of complexity. “Chinese open-weight models like dots.llm1 may be technically transparent—but in the eyes of global enterprises, transparency is no substitute for trust. Especially in regulated industries — banking, healthcare, and defense — geopolitical risk is now baked into AI architecture decisions,” Gogia warned.

What this means for business

The rise of capable, free AI models from Chinese companies is forcing a rethink of enterprise AI strategy. Companies that once assumed they’d need to pay premium prices for cutting-edge AI now have alternatives that might work just as well for many tasks, but with new considerations around governance, geopolitics, and long-term strategy.

RedNote’s entry into open-source AI represents more than technological competition — it’s part of a fundamental shift in how AI power will be distributed globally. While these models offer compelling technical capabilities at attractive price points, enterprises must weigh transparency benefits against new governance challenges and geopolitical considerations.

Five men from China, the United States, and Turkey pleaded guilty to their involvement in an international crime ring and laundering nearly $37 million stolen from U.S. victims in cryptocurrency investment scams carried out from Cambodia. [...]

Modern enterprise networks are highly complex environments that rely on hundreds of apps and infrastructure services. These systems need to interact securely and efficiently without constant human oversight, which is where non-human identities (NHIs) come in. NHIs — including application secrets, API keys, service accounts, and OAuth tokens — have exploded in recent years, thanks to an [email protected]

Google has stepped in to address a security flaw that could have made it possible to brute-force an account's recovery phone number, potentially exposing them to privacy and security risks. The issue, according to Singaporean security researcher "brutecat," leverages an issue in the company's account recovery feature. That said, exploiting the vulnerability hinges on several moving parts, Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]

The Arkana Security extortion gang briefly listed over the weekend what appeared to be newly stolen Ticketmaster data but is instead the data stolen during the 2024 Snowflake data theft attacks. [...]

While the new user interface design Apple execs highlighted at this year’s Worldwide Developers Conference (WWDC) might have been a bit of an eye-candy distraction, enterprise users were not forgotten, with announcements unveiled that could make a difference in the business world.

Apple’s WWDC keynote speech was far from the disappointment many had anticipated, given the number of changes announced. In addition to that new user interface, which will show up across all Apple products and is called Liquid Glass (we’ll look at it in more depth elsewhere), there were also APIs designed to help developers work AI into their software. The last is a nod to the company’s slow efforts to make Apple Intelligence the showcase tech promised at last year’s WWDC.

But for now I’m focused on just some of the changes Apple introduced that should have the most impact on enterprise users. Many, but not all, of these changes will be similar across all Apple’s devices, whether that’s macOS, iOS iPadOS, or others.

What’s a computer? The iPad

For me, the big news concerns the iPad, which gains noteworthy improvements that will transform its use. Not only can you open multiple app windows at once, but Apple has done this with a clever and versatile windowing system that even supports windows resizing, pervasive position memory and more. (You can even see all your currently open windows in Expose.)

These multitasking and windowing improvements should make it much easier to engage in professional work on an iPad and are the kind of improvements people have wanted for years. Changes on the iPad extend to the Files app, too, which gets important updates that will make it particularly useful.

Files for the rest of us

All Apple users, not just iPad owners, should see benefits from changes coming to Files. For iPad users, they make it easier to find and work with the files you need, and to add additional sorting options in the Files window; but Files is an app, which implies that at least some of those improvements will be available across other Apple devices. Pro users will likely also appreciate the new ability to customize folders with symbols, emoji, and colors.

The transformation of automation

Apple rolled out improvements in Shortcuts, including the capacity to create them and the ability to use Spotlight to identify and invoke them. And since the Shortcuts you do create can sync across all your devices, users will be able to build powerful productivity shortcuts on one device and use them on another, where they’re supported. This integration is, of course, also reflected by the Liquid Glass UI, which should create a real sense of familiarity as you move from Mac to tablet to iPhone.

Spotlight is evolving

Apple also announced big changes in Spotlight search on the Mac, including intelligent actions built around Shortcuts and Apple Intelligence, as well as Spotlight actions and quick keys; the latter automatically surface actions you might want to take with a Spotlight-selected item and commit those actions from within the search. This is powerful — you can find an item, change and save it, and share it, all from within Spotlight. The software can also learn from what you do, offering personalized actions for what you might want to do. 

“During a search, all results — including files, folders, events, apps, messages, and more — are now listed together and ranked intelligently based on relevance to the user,” Apple said. N”ew filtering options rapidly narrow searches to exactly what a user is looking for, like PDFs or Mail messages. Spotlight can also surface results for documents stored on third-party cloud drives. And when a user doesn’t know exactly what they’re searching for, Spotlight’s new browse views make it easy to scan through their apps, files, clipboard history, and more.”

You can also take hundreds of actions from directly inside Spotlight, such as sending an email or writing a note. While many of these features are being promoted as for the Mac, Spotlight operates across Apple’s platforms, so it will be interesting to see where else these features might show up as beta testing begins and future support rolls out.

App Intents

That brings me to App Intents. App Intents have been around for a while, but the framework will now let developers make actions in their apps in available across the system via Shortcuts or Spotlight. If nothing else this, should make work itself easier to get on with, and as those improvements percolate across Apple’s ecosystem, the hardest task will probably become simply remembering what actions you are able to take.

App Intents also gain support for visual intelligence, which means apps will be able to provide and make use of visual search results.

Etsy CTO Rafe Colburn explained what this means in a statement provided by Apple: “The ability to meet shoppers right on their iPhone with visual intelligence is a meaningful unlock and makes it easier than ever for buyers to quickly discover exactly what they’re looking for while directly supporting small businesses,” he said.

Vision Pro

Enterprises are already using Vision Pro devices, and Apple talked up the spatial computing headset during its WWDC keynote. Given the cost of these devices ($3,499), it is a welcome change that it’s now much easier to share them in workgroups; each user can save their own vision OS profile to their iPhone and then set the vision device up to suit. Apple also unveiled a new Vision Pro tool that lets you save a piece of content as Protected Content. (There are new APIs app developers can use for this, too.) The intention here is to ensure content shared on a Vision Pro doesn’t get shared outside your business.

There’s also support for a new Vision Pro peripheral, one that doesn’t come from Apple. Logitech Muse is a new pen/control device for spatial reality that lets you tweak and draw when working on projects with others. Look to Scroll allows users to explore apps and websites using just their eyes, and users can customize the scroll speed to their liking. Not surprisingly, developers will be able to integrate Look to Scroll into their visionOS apps.

Apple Intelligence was everywhere

Reading between the lines, you can see that artificial intelligence is actually everywhere in this year’s WWDC announcements. One of the biggest announcements does hint at the contextual intelligence Apple has previously told us about: Visual Intelligence now understands what is on screen, lets you ask questions about what you are looking at, and allows you to search for specific items or add events to your calendar. Apple has, of course, made it possible for developers to build support for this feature within their apps. 

“Last year, we took the first steps on a journey to bring users intelligence that’s helpful, relevant, easy to use, and right where users need it, all while protecting their privacy. Now, the models that power Apple Intelligence are becoming more capable and efficient, and we’re integrating features in even more places across each of our operating systems,” said Craig Federighi, Apple’s senior vice president of software engineering.

What this means: developers can now use Apple Intelligence APIs to build its tools within their apps, thanks to the Foundation Models Framework.

Developers can also look forward to Xcode 26, which can connect large language models (LLMs) directly into their coding experience, thanks to built-in support for ChatGPT. (Developers can also use API keys from other providers or run local models to support Xcode.)

Apple Intelligence features will be coming to eight more languages by the end of the year: Danish, Dutch, Norwegian, Portuguese (Portugal), Swedish, Turkish, Chinese (traditional), and Vietnamese. 

Live Translation

Live Translation is profound. It’s built into Messages, FaceTime, and iPhone and runs entirely on device. In use, it can automatically translate messages between languages, offer translated live captions on FaceTime calls, or speak the translation aloud during a conversation while on a phone call. It’s a fantastic feature I can’t wait to try and I expect to see it used in very interesting ways. 

Apple also did confirm a new system to help promote accessibility in apps. Accessibility Nutrition Labels for App Store product pages will let developers tell customers what accessibility features their apps support.

Summing up, even if this is a lean year for flashy announcements, the size of Apple’s ecosystem is now so vast the company is still capable of introducing changes that will delight customers, while also helping them get things done.

Apple rolled out the first betas of the new operating systems on Monday, with public betas following over the next few weeks. So you will be able to see for yourself which improvements make the most difference to you. Follow me through one of the networks below to learn more about these changes as I explore them.

You can follow me on social media! Join me on BlueSky,  LinkedIn, and Mastodon.

Sluggish sales of Apple’s Vision Pro mixed reality headset haven’t dampened the company’s enthusiasm for advancing the device’s 3D computing experience, which now incorporates AI to deliver richer context and experiences.

Apple has upgraded the newly renamed visionOS 26 operating system to improve usability, blending AI into the background to enhance images and user interaction, Apple executives said at the company’s Worldwide Developer Conference on Monday.

Apple calls Vision Pro a “spatial computer” in which users can watch movies or interact with digital experiences in the real world using hands, eyes, and other senses.

A new feature called “spatial scenes” uses AI to give a 3D feel to traditional two-dimensional photos, which will be available in the Photos app. The AI model uses computational depth to add more perspectives to images, Haley Allen, senior director of visionOS program management, said during the presentation.

“Spatial scenes will make your web browsing experience more engaging, too,” Allen said.

In Safari, users will be able to select a feature called “spatial browsing” to view two-dimensional inline photos in supported articles in a three-dimensional format. Apple also introduced a curated app called “spatial gallery” to showcase curated spatial image content.

And the company is making major AI-driven changes to its Persona feature, which will now create more realistic avatars of users for the FaceTime video chat application. Previous avatars had lifelike renditions, but the updated Personas will be more personal and familiar, Allen said.

“Hair, lashes, complexion, all look remarkably accurate,” Allen said.

Apple also added many AI-driven improvements to apps such as Image Playground, which are supported across Apple devices. Apple has updated Image Playground to create more creative emojis by tapping into OpenAI’s ChatGPT AI application, which supports text and image creation.

The company also introduced a variety of non-AI features.

Vision Pro users will now be able to access widgets using their headset. The widgets will remain fixed within the visual field and will be clearly accessible at all times, Mike Rockwell, vice president of the Vision Products Group, said during a pre-recorded Apple video detailing the new features.

Widgets such as calendar and battery indicator are already popular in macOS. The company also introduced new widgets, such as a circular clock, music, real-time weather, and photos that can show snapshots of pictures from personal collections.

The widgets are blended into a user interface built around the new “liquid design” color scheme, which focuses on sharp colors and window edges.

Users can also now share Vision Pro experiences with others, which could help enterprises collaborate on three-dimensional engineering and design applications. Teams in hundreds of companies already use Vision Pro headsets for design, sales, training, and more, Rockwell said.

With that in mind, companies can also now create a pool of Vision Pro devices that can be shared among team members and guest users. Users can save their hand, eye, accessibility, and prescription information to their iPhone and activate a shared headset.

Apple also added new APIs such as “Protected Content,” which can show confidential materials to authorized users.

The company is also adding new input options such as the Logitech Muse, a pen-like device that can be used to draw within the 3D interface, and support for the Sony PlayStation VR2 Sense Controller; the latter can be worn on the hand to play motion-controlled video games within the headset.

Over 84,000 instances of the Roundcube webmail software are vulnerable to CVE-2025-49113, a critical remote code execution (RCE) vulnerability with a publicly available exploit. [...]

A vulnerability allowed researchers to brute-force any Google account's recovery phone number simply by knowing a their profile name and an easily retrieved partial phone number, creating a massive risk for phishing and SIM-swapping attacks. [...]

SentinelOne has shared more details on an attempted supply chain attack by Chinese hackers through an IT services and logistics firm that manages hardware logistics for the cybersecurity firm. [...]

AI research organization Eleuther AI has launched a massive text database, Common Pile v0.1, that can be used to train AI systems, according to Techcrunch. The 8TB database consists exclusively of publicly licensed texts, or texts that are classified as public domain.

Common Pile v0.1 was developed over two years in collaboration with Poolside, Hugging Face, the US Library of Congress and the University of Toronto, among others.

The data collection was released after concerns arose about several generative AI (genAI) companies using copyrighted material to train their models without the permission of the copyright owners. Eleuther AI was also behind the collection, The Pile, which has become a central point in the debate; it now wants to show with Common Pile v0.1 that training is possible without copyrighted material.

Common Pile v.01 was reportedly used to train the Comma v0.1-1T and Comma v0.1-2T AI models; Eluther AI claims Comma v0.1-2T performs as well as Meta’s first Llama model in terms of programming, image understanding and math. Eluther AI plans is release more open data collections in the future.

A collective of former WordPress developers and contributors backed by the Linux Foundation has launched the FAIR Package Manager, a new and independent distribution system for trusted WordPress plugins and themes. [...]

Filling the void in the few hours before WWDC begins, Apple’s machine learning team raced out of the gate with a report to make people think twice about artificial intelligence, arguing that while the intelligence is artificial, it’s only superficially smart.

Some seem to think that Apple is attempting to mask its slow progress in AI development as its competitors push ahead toward artificial general intelligence (AGI). Others warn that perhaps we should all think about these findings, given the speed with which the technology is being deployed across almost every part of our society. Are we really becoming dependent on tech that doesn’t really work and will simply repeat the prejudices of its owners?

Because that’s what you get when you use something that is basically the kind of glorified pattern matching Apple’s researchers reveal in their latest paper, “The Illusion of Thinking.”

The Illusion of Thinking

In the 32-page paper, Apple’s machine learning experts argue that despite the impressive results they can achieve, AI models aren’t actually capable of authentic reasoning, relying more on pattern matching than deduction. For its analysis, the team tested models from OpenAI, Google DeepMind, Anthropic, and DeepSeek.

“Recent generations of frontier language models have introduced Large Reasoning Models (LRMs) that generate detailed thinking processes before providing answers,” they write.“Through extensive experimentation across diverse puzzles, we show that frontier LRMs face a complete accuracy collapse beyond certain complexities. Moreover, they exhibit a counterintuitive scaling limit: their reasoning effort increases with problem complexity up to a point, then declines despite having an adequate token budget.”

They also claim that while both LRM and LLM systems are good at some things, LLM wins at low-complexity tasks, LRMs do well in medium complex tasks, and both LLM and LRM models “experience complete collapse” when handling high-complexity tasks.

“We found that LRMs have limitations in exact computation: they fail to use explicit algorithms and reason inconsistently across puzzles. We also investigate the reasoning traces in more depth, studying the patterns of explored solutions and analyzing the models’ computational behavior, shedding light on their strengths, limitations, and ultimately raising crucial questions about their true reasoning capabilities,” the researchers wrote.

It raises the critical question: are the genAI systems we think are capable of reasoning actually doing so, or have they simply been taught to appear to be reasoning, even when they aren’t.

Between AI ideas and reality, falls the shadow

To prove its point, Apple shows how existing systems can be fed irrelevant or fake data they don’t recognize, leading to failures. The inference is that while these machines are good at repeating and identifying the patterns they know, actual creative problem-solving is beyond their abilities.

AI struggles with abstraction and generalization, which means the notion we’re anywhere near AGI…, well, it’s not going to happen anytime soon.

Apple isn’t alone in warning about resisting the hype around AI. Critic Gary Marcus has been raising red flags about the true intelligence in these systems, pointing to their over-reliance on trading data. This reliance means that while we’re promised that genAI will change everything and make everything better and more productive, it probably won’t achieve anything like those promises. The gulf between promise and reality is driving many in the industry to reassess their expectations already. 

What this means, of course, is that rather than becoming reliant on these systems, it makes more sense to remain pragmatic and develop a hybrid approach, both toward AI deployment AI development. 

Hold the door

Apple would say this, I suppose. We know it is facing unexpected challenges realizing the promises it made for Apple Intelligence at WWDC last year. “We need more time to complete our work on these features, so they meet our high quality bar,” Apple CEO Tim Cook said during last month’s earnings call. “We are making progress, and we look forward to getting these features into customers’ hands.”

It’s not impossible that part of the reason Apple has delayed introduction of some AI features is quite simply because they don’t work as they should. It’s possible this reflects what Apple’s research shows, with the abilities of Apple Intelligence constrained by the same challenges its report describes. 

To some degree, Apple Intelligence doesn’t matter. What is important is that any move to bet the entire future of you or of your business on AI is doomed to fail right now, if only because at root even automated pattern matching will eventually repeat the errors of the past. In the future? Who knows — the ability to create machines with human level intelligence is evidently in process, though most humans will probably prove more a little resistant to being replaced by machines. 

In the meantime, Apple will bring us a rebranded Siri, APIs so developers can build Apple Intelligence within their apps, and more partnerships with AI providers; the final obstacle is whether Apple can evolve an iteration of AI that gives it a truly competitive edge, and how long the company can play for time to achieve that.

You can follow me on social media! Join me on BlueSky,  LinkedIn, and Mastodon.