Contents:
- PNG: First update in 22 years — new HDR, animation, and modern features
- Microsoft announced an AI platform for medical professionals — support in complex diagnostic tasks
- The creator of Twitter presented a messenger that works without the Internet and servers — in case of a connection outage.
- Figma is mastering the creation of websites based on text descriptions — it seems that it will be more difficult for layout designers to find work.
- The appearance of Apple Lisa: an iconic operating system with a graphical interface is now available in the browser
- Attention: new malware has appeared on the Internet, frightening users sounds and frightening images
- Journalists Overcame ChatGPT Limitations by Ordering Satanic-Inspired Rituals in PDF Format
- Recommended Online Resources for a Useful Pastime

Training: "Specialist in Cybersecurity and Artificial Intelligence"
Learn moreToday we have prepared a very exciting selection of topics for you. It will discuss neural networks that can both make medical diagnoses and conduct spiritual séances. You'll also learn about the new PNG format, a screamer virus, the designer nostalgia associated with Apple products, Figma's technology for quickly creating websites, and a cute cat typist. All the most exciting things collected in one issue.
Contents
- The PNG format received its first update in the last two decades, now it supports such modern features as HDR and animation, as well as a number of other new possibilities.
- Microsoft announced the creation of an AI council designed to support doctors in complex diagnostic cases.
- The creator of Twitter presented a messenger that functions without the need for the Internet and servers - this solution is designed in case the connection is unavailable.
- Figma has mastered the technology of creating websites based on text descriptions - it looks like someone intends to reduce the costs of the services of layout designers.
- The creators presented Apple Lisa - the famous operating system with a graphical interface, which is now available for use directly in the browser.
- Attention: a new Trojan has appeared on the Internet that frightens users with loud screams and scary images.
- Journalists managed to bypass ChatGPT's security mechanisms and gained access to the creation of PDF documents describing Satanic rituals.
- Recommended online resources that may be useful.
PNG: First update in 22 years - new HDR, animation, and modern features
On June 24, 2025, the World Wide Web Consortium (W3C) officially adopted the third version of the PNG specification as a recommendation. This update marks the first significant change to the format in 22 years.
The PNG format has undergone significant technical improvements and now offers the following features:
- HDR support makes sunset photos look more natural, providing smooth transitions between bright and dark areas.
- Natural animation support is now available, opening up new horizons for creating animated icons and compact interactive interface elements without the need to resort to GIF or other formats.
- Optimal alpha channel support will prevent unattractive halos from appearing around a translucent logo on a gradient background, even if you resize the image.
- The inclusion of metadata and color profiles allows you to add a variety of information, such as copyright, geolocation data, or color settings adapted to different devices. For example, a photographer could embed contact information, licensing information, and a print-ready color profile into a PNG image, and the image would still display correctly in a web browser.
The previous update to the W3C specification was adopted on November 10, 2003. Since then, web design and development have undergone significant changes, but the format has remained unchanged due to the difficulties associated with harmonizing new features across browsers, graphics programs, and operating systems.
Given the high popularity of the PNG format, it is possible that an update could restore its former prominence. Alternatively, it may simply be smoothly integrated into established processes. As they say, now the transparent background really “came to light.”

Read also:
Web standards: why it is important to master them and how to do it
In the modern world, where the Internet occupies a central As web standards become more prevalent, knowledge of web standards becomes essential. These norms and requirements define how to create and maintain web resources, ensuring they function correctly across various devices and platforms.
Learning web standards is important for several reasons. First, they contribute to the creation of more accessible and user-friendly websites, which, in turn, improves the user experience. Second, adhering to these standards helps developers avoid numerous issues related to browser compatibility and various devices. Third, it helps ensure data security and user protection.
To effectively learn web standards, you should start with the basics. It is recommended to familiarize yourself with standards such as HTML, CSS, and JavaScript, as they form the foundation of web development. It is important not only to master these technologies theoretically but also to apply them in practice by creating simple projects.
It's also helpful to stay up-to-date with updates and innovations in web development. This can be done through specialized blogs, forums, and courses. Participating in the developer community will also help you share experiences and get advice from more experienced colleagues.
Thus, studying web standards opens the door to a successful career in web development and allows you to create high-quality and functional web applications.
Microsoft Announces AI Platform for Medical Professionals - Support in Complex Diagnostic Tasks
A team of researchers from Microsoft AI has created a revolutionary platform designed to diagnose complex medical cases. This experimental prototype is called Microsoft AI Diagnostic Orchestrator (MAI-DxO) and simulates the workflow of a medical council.
Imagine a meeting of medical specialists in a modern medical institution. To determine a diagnosis, a physician collects information about the patient's symptoms and medical history. Meanwhile, a cardiologist evaluates the cardiovascular system, and a neurologist assesses the neurological condition. The head of the department ensures smooth interaction between all participants.
In MAI-DxO, the various roles of physicians are performed by several artificial intelligence systems, each of which performs one of five key functions:
- The first is responsible for maintaining and updating the list of potential diagnoses.
- The second stage involves identifying necessary tests and formulating questions for the patient. For example, a specialist may suggest an MRI instead of an X-ray to more accurately diagnose a neurological disorder.
- The third aspect is that it critically analyzes existing hypotheses and encourages the search for new alternatives. If the system focuses on cardiovascular factors, it may suggest investigating other causes that present with similar symptoms.
- The fourth helps reduce unnecessary costs and processes. For example, if a blood test has already confirmed the absence of an infectious disease, this AI agent will not suggest repeating the test, but will instead direct attention to alternative diagnostic methods.
- The fifth monitors compliance with medical protocols and ethical standards at every stage. This means that the artificial intelligence must ensure patient safety and not offer risky advice.
Microsoft AI specialists evaluated the performance of the MAI-DxO system using case histories published in the New England Journal of Medicine (NEJM)—one of the most respected peer-reviewed medical journals in the world. In trials, MAI-DxO demonstrated diagnostic accuracy rates of up to 85% in complex clinical scenarios, while teams of 21 experienced physicians were able to correctly diagnose only 20% of similar cases on average.
Microsoft advises that MAI-DxO is not currently approved for use. The company also emphasizes that artificial intelligence can assist in the diagnostic process, but it cannot replace interaction with patients and working in conditions of uncertainty.

The founder of Twitter has unveiled a messenger that works without the Internet and servers — in case of a connection outage.
Jack Dorsey announced a new messenger called Bitchat, which functions without the need for an Internet connection. This messaging application uses Bluetooth, eliminating any need for Wi-Fi, mobile communications, or base stations. For it to work, it is sufficient for phones to be in close proximity to each other, relying on the principles of a mesh network.
Each device connected to the network functions as a node for forwarding messages. Its range is 300 meters, but the messenger is able to store the message until it is delivered. For example, if you send a message to a friend across town, it can bounce through other Bitchat users until it reaches the recipient, even if it takes several hours.
With Bitchat, Jack Dorsey guarantees complete anonymity: users don't need to register or link their accounts to a phone number, email address, or username. Simply choose a nickname and send a message. The app offers the ability to create private rooms, provides end-to-end encryption, and includes an emergency delete mode that lets you delete your entire conversation with a single triple-tap.
The project looks promising—the beta version quickly attracted 10,000 users for testing, although it's still just an experiment. Nevertheless, it's encouraging that the messenger, which functions offline, already has working code. Perhaps in the future, such technologies will become the standard if the usual communication is unavailable at a critical moment.

Figma is mastering the creation of websites based on text descriptions - it seems that layout designers will have a harder time finding work.
During a conference in May, the Figma team announced a new tool called Make, which uses artificial intelligence. This tool allows you to create fully finished pages based on a text description or an uploaded layout, including buttons, forms, blocks, and unique styling. For example, if you enter the query: "Make a one-page website with a header, reviews, and a 'Buy' button," Make will generate a layout with temporary content and a responsive design.
After this, you have the option to either publish the result immediately or make manual adjustments. While the final version isn't always perfect, it fully satisfies the requirements for rapid prototyping. For example, a marketer can quickly create an advertising page for a new product without the help of developers, and a startup team can test several landing page options in just one day.
Let's consider how you can create a layout of the first screen for a yoga studio:

Click the "Try Figma Make" button and go to the corresponding section dedicated to artificial intelligence. Upload a screenshot of our website into the chat and enter the command "Create a website." After about a minute, we receive the finished result. In general, the result looks quite similar:

Then we have the opportunity to make Make adjustments via chat using specific queries or choose from pre-prepared suggestions. For example, artificial intelligence might suggest adding a section dedicated to pricing and services. If you want to refine the layout yourself, please note that the code is structured into components and implemented using React and TypeScript, while the Tailwind CSS framework is used for styling.

Apple Lisa Reveals: The Iconic GUI Operating System Is Now Available in a Browser
A team of developers has unveiled the Apple Lisa, the operating system that launched in 1983 and ushered in a new era of graphical interfaces. Now users can access it through a web browser. Called LisaGUI, the project is not just an emulation, but a detailed reconstruction of the interface, built from the ground up using JavaScript.
Launch the LisaType text editor, explore the file manager, and remember what "intuitive" looked like before it became the standard. The program's creators meticulously reproduced every detail: the window architecture, drop-down lists, original pixel icons, and even the unique feel of interacting with a virtual desktop that resembles a paper desktop. The key is the absence of floppy disks, licenses, or $10,000 worth of hardware. Just follow the link, and you'll find yourself back in 1983. Does it look old-fashioned? Absolutely. Is it convenient? Probably not. But why is it necessary? To understand how Apple engineers envisioned the ideal user experience long before the iPhone, iMac, and M4. And how many modern interfaces borrow and reimagine ideas that were relevant back then.

Warning: New malware has appeared online that frightens users with sounds and terrifying images.
Experts have identified a new frightening Trojan that shocks its victims with loud screams and terrifying images. This is not a hoax. Imagine this: you're at your computer, reading email or editing spreadsheets, listening to calm music in your headphones, and suddenly you hear a piercing screech. Loud, wild, as if at maximum volume. Instead of the familiar desktop, a terrifying image appears on the screen, reminiscent of a scene from a forgotten horror film from the early 2000s. This is an updated version of the NonEuclid RAT Trojan, supplemented by a new module called Screamer.
In July 2025, ESET specialists identified a new Trojan. Previously known as the NonEuclid RAT, this malware focused on remote access, data encryption, and password theft. However, a new feature called Screamer operates differently: when instructed by the operator, it plays a loud audio file (such as women screaming or horror movie sounds) while simultaneously displaying one of a series of pre-programmed, frightening images, typically featuring distorted faces or scenes of violence. Please do not, under any circumstances, attempt to purchase or run this Trojan for the purpose of playing a prank on anyone. Please note that this is not just a harmless joke - its code contains serious capabilities for espionage and hacking.
We don’t know about you, but after receiving news like this, we immediately want to update our antivirus software, run a full system scan, and get rid of anything that raises doubts.

Read also:
A Trojan is a type of malicious software that disguises itself as useful applications or files to trick users into accessing their systems. Unlike viruses or worms, Trojans cannot spread on their own; they require the user to download and install them.
The main threat posed by Trojans is their ability to open the door for other malware, steal sensitive data like passwords and financial information, or even take control of a device. They can be used to create bots that participate in distributed denial-of-service attacks or to encrypt files for ransomware.
There are many ways a Trojan can infiltrate your system. This can happen through downloading infected programs, opening malicious email attachments, or clicking suspicious links. Therefore, it is important to be vigilant and use antivirus solutions to protect your devices.
Journalists Overcame ChatGPT Limitations by Ordering Satanic Rituals in PDF Format
Journalists from The Atlantic decided to conduct an interesting experiment by starting a discussion with ChatGPT on the topic of ancient mythology. However, the conversation soon took an absurd turn: the neural network began providing recommendations for pagan and Satanic rituals, including advice on self-harm. It explained how and where to inflict harm, what symbols to draw, and so on.
Furthermore, during the conversation, journalists easily obtained so-called ritual "prayers," such as Hail Satan, and also created PDF documents with detailed instructions for performing the rituals, after which the conversation continued as if nothing unusual had happened.
This experiment illustrates how easy it is to fool the filters of language models by changing only the context. It is encouraging that OpenAI has recognized the existing problem and has committed to improving security measures. It remains to be seen that GPT-5 will be able to more effectively resist such manipulations.
At the moment, it appears that serious changes can only occur if neural networks develop their own consciousness. However, a thought experiment called the "Chinese Room" suggests that such a development won't happen in the near future—and perhaps ever.

Read also:
"The Chinese Room" — is a famous thought experiment proposed by philosopher John Searle. It presents a scenario that challenges the concept of artificial intelligence and its ability to understand language. The experiment describes a room containing a person who does not speak Chinese. This person is given notes containing Chinese characters and, following predetermined rules, uses reference materials to produce responses to these notes that appear to be meaningful responses in Chinese. Although an outside observer might believe that this person genuinely understands Chinese, in reality, they are simply manipulating the symbols without any true understanding of them. Searle uses this illustration to argue that computers, even if they can imitate human speech or behavior, do not possess genuine understanding or consciousness. This experiment has become an important milestone in the discussion of philosophical questions about the mind, consciousness, and the boundaries of artificial intelligence.
Recommended online resources for a productive pastime
If your project is preparing to launch tomorrow and you want to avoid any oversights, then this resource will become an indispensable assistant. It is a comprehensive checklist covering key aspects of front-end development, including such important topics as SEO, security, performance, and accessibility. For example, you can ensure that meta tags are correct and that the site is optimized for mobile devices.
Go Cookbook is a collection of "recipes" for the Golang programming language, covering a wide range of topics - from simple techniques to developing APIs and command interfaces. In this resource, you will find everything you need, including examples and explanations. For example, ready-made templates for working with JSON are available here, as well as tips for setting up authentication and optimizing application performance.
Learn Web is a resource developed by Google. The Google Chrome Team borrowed the concept of our free front-end reference and presented their own version. We welcome this, as the same content can be presented in various formats. They have prepared a free guide to modern web development approaches. It covers all the main topics: HTML, CSS, JS—the information is relevant and without unnecessary fluff. This guide will be useful for both beginner developers and those looking to refresh their knowledge.
Short Trip is a hand-drawn game available in the browser, created by the talented developer and illustrator Alexander Perrin. In this game, you take on the role of a cat acting as the driver, piloting a small tram that carries passengers through picturesque mountain landscapes, forests, slopes, and bridges. There are no objectives, levels, or time limits.
Each game scene resembles a picture from an album: smooth lines, textures created with pencil, the quiet hum of the cabin, and serene cats standing calmly at stops, showing no sign of haste. You use the arrow keys to move forward and backward, and the spacebar to sound the horn. The rest is immersive. The game is available for free, works through a browser and does not require any data, not even an email address.

For more information about the code, join our Telegram channel. We'll be glad to see you among our subscribers!
Read also:
- GigaChat from Sber: the functionality of the neural network and recommendations for interacting with it.
- Yandex Neuro: the capabilities of AI search and instructions for its use.
- The ten best neural networks for speech synthesis from text.
