🎮 Roblox Aims to Turn into an AI Game

Roblox has acquired Morpheus AI, a developer of "world models" designed to generate interactive 3D worlds in real time.

Additionally, the founders of Dynamics Lab and Lucid AI—two other prominent startups in the "world model" field—have previously joined Roblox.

All the acquired teams will work on a unified Roblox Reality architecture. It will combine real-time AI generation with the existing game engine, keeping the logic and multiplayer capabilities intact.

The ultimate goal is to achieve stable AI-generated game scenes at 4K and 60 Hz.

Would you play this kind of "AI-Roblox"?

❤️ — Yes, that's a dream!
🔥 — No, it's going to be trash

@hiaimediaen

🚀 China Developed a Washing Machine for Astronauts—No Water, No Detergent

In space, water is scarce. Every drop is recycled and used for drinking and hygiene. So astronauts don't wash their clothes—they wear them until they can't anymore, then send them to burn up in the atmosphere along with cargo capsules.

But engineers at the China Astronaut Research and Training Centre have recently developed a compact washing machine for use in microgravity.

💻 The device is slightly larger than a suitcase and weighs 12 kg. In a single cycle, it uses just 400 ml of water delivered as an ultra-fine mist and cleans up to 800 grams of clothing. Instead of detergent, the machine disinfects clothes with ozone produced by ultraviolet light.

Simulations show a sterilization rate of 99.9%. One cycle protects clothing for up to five uses. The device is designed to last five years.

➡️ The technology could potentially reduce the amount of clothing needed for space missions by more than 60%—significantly cutting the weight of cargo at launch.

The developers are confident that this kind of washing machine will be essential for missions to the Moon and Mars.

Did you know astronauts don't wash their clothes?

❤️ — Yes, obviously
🔥 — No, that surprised me!

@hiaimediaen

🎥 C++: The Documentary—a Film About the Language That Powers Half the World

A documentary film C++: The Documentary by CultRepo has dropped on YouTube.

❓ In 2025, around 16 million developers were writing in C++, roughly one in three worldwide. The language powers games, browsers, databases, blockchain protocols, financial systems, and even the physics calculations at CERN. Many key AI libraries are written in C++ too.

At the center of the film is the story of Bjarne Stroustrup, who started writing C with Classes at Bell Labs in 1979. The film also features Herb Sutter, Alexander Stepanov, and dozens of other key contributors to C++.

🤖 The film is full of vivid details and memories. In the mid-1980s, ahead of the language's commercial release, AT&T gave Stroustrup $5,000 over three years to promote the language. He spent $3,000 of it on a party at the first C++ conference — and the language quickly became popular anyway, thanks to its speed and versatility.

"I've launched this thing on the world and I feel I'm obliged to help make it better," says Stroustrup.

➡️ Watch the film here.

Have you ever programmed in C++?

❤️ — Yes, it's classics!
🔥 — No, never got into it...

@hiaimediaen

💡 Nvidia CEO Dreams of Dying on the Job

In an interview with CNA, Jensen Huang said he wants to work as long as possible, and dying on the job would be "a dream come true."

He can't imagine a more meaningful life. On top of that, both of his children work at Nvidia—he gets to see them every day and couldn't be prouder.

💰 In another recent appearance, Huang said Taiwanese parents are tough on their kids. "I'm 63, and they still think I'm not good enough," says Huang, ranked #8 on the Forbes list of the world's wealthiest people, with a net worth of around $180B.

🔄 And yet, he runs his own company much the same way: always critical of everyone's work, so he can help them get better.

What do you think of this leadership style?

❤️ — Huang is a true genius
🔥 — Feels like too much...

@hiaimediaen

⏫ Canadian Engineers Built an "Unbeatable" AI Air Hockey Robot

A team of engineers from the University of British Columbia spent about 1.5 years on the project.

Air hockey looks simple, but teaching a robot to play it is anything but. The puck moves fast and unpredictably, and even the tiniest surface imperfections or uneven rails throw off its trajectory. That makes traditional physics engines practically useless for real-time calculations.

So the developers trained the robot entirely in simulation, carefully accounting for every possible real-world imperfection. The AI played thousands of matches in parallel against opponent agents. The wooden rail bounces turned out to be so chaotic that the team trained a separate AI-model just to predict them.

🤖 In the end, the "AI brain" loaded into a physical robot turned out to be practically unbeatable for an average player.

Would you play against this robot?

❤️ — Yes, sounds fun!
🤔 — What's the point...

@hiaimediaen

🚀 Apple Makes Siri a Real AI Assistant

At WWDC 2026, Apple announced the biggest update to Siri. The AI in the Apple ecosystem now runs on a mix of the company's own models and Google Gemini technology. The assistant is now officially called Siri AI.

Key Features:

✅ Siri "lives" in the Dynamic Island. It gives suggestions right in notifications, can summarize messages, or show directions.

✅ Siri AI now has its own standalone app. All conversations are available across all devices, including Apple Watch.

✅ It's a true agent. The AI can analyze the screen, create notes or reminders during a conversation, remember details, and adapt to the user.

✅ The assistant understands any screenshot. It can explain what's on it or turn it into a note.

✅ On macOS, the AI is built directly into Spotlight. The assistant can search and analyze files.

✅ AI is added to system apps: sorting tabs in Safari, working with passwords, an AI photo editor, a smart calendar, and much more.

✅ Siri will even work in the Camera app. For example, you take a photo of a receipt and the AI splits the bill among friends. Or it suggests the best angle for your shot.

✅ Siri can write full documents or letters. And the AI will adapt to your personal style when talking to different people: friends, colleagues, or family.

🔍 Developer access opens today. User features will roll out later this year.

Waiting for the new Siri?

❤️ — Yes, can't wait!
🔥 — No, too much AI

@hiaimediaen

❗️ Scientists Use Precise Gene Editing in Human Embryos for the First Time

Researchers used base-editing technology to modify three genes in human embryos at an early stage of development. In theory, this could make it easier to correct harmful genetic mutations before they affect the developing organism.

The team changed individual DNA "letters" in PCSK9, which regulates "bad" cholesterol levels, HBG1 and HBG2, which are linked to inherited blood disorders.

⚙️ Base editing is a newer, more precise version of CRISPR-Cas9. Unlike the traditional approach, which cuts the DNA strand and relies on the cell to repair it, base editing directly swaps one DNA letter for another without creating a break, reducing the risk of unwanted errors.

The researchers successfully introduced the desired edits in all three genes. However, the results were far from perfect: some cells carried the edited version, while others remained unchanged. The team acknowledges that the technology still needs significant improvement.

❓ Still, the work has already sparked debate within the scientific community. Some researchers see it as a "conceptual shift" that moves the field forward in a careful and ethical way. Others worry that the technology could eventually encourage wealthy parents to pursue "designer babies" engineered for traits such as intelligence, appearance, or athletic ability.

Human embryos were previously edited in 2018 by He Jiankui, who used CRISPR-Cas9 in an attempt to make embryos resistant to HIV. The experiment resulted in the birth of twin girls, but it was widely condemned as unethical and unsafe. He was sentenced to three years in prison.

What do you think about genetically modified babies?

❤️ — It's the future of medicine!
🔥 — It's far too risky...

@hiaimediaen

🔍 Gemini Watermarks Found in War Thunder Textures

War Thunder players spotted Gemini watermarks on several in-game assets sold for real money. Gaijin Entertainment has acknowledged the issue, removed the disputed elements, and launched an internal review of its contractors.

According to the studio, an outsourced team used AI tools to process references and draft materials without approval. Gaijin says this violated its internal rules, which ban contractors from using generative AI, and the issue slipped through quality control before release.

The affected assets included camouflage for the M60 tank and decorative elements in the Challenger 2 OES pack. Gaijin rejected a separate claim about the Yak-130 model: the star that looked like a Gemini watermark turned out to be a real technical marking from aircraft reference photos.

Are AI-generated textures in games acceptable?

❤️ — Yes, if the result is good
🔥 — No, it is just a way to save money on artists

@hiaimediaen

🌥 Claude Fable 5: The Smartest and Terribly Human Model from Anthropic

Anthropic has released Fable 5, the first model in their new line of super-powerful neural networks.

It is based on the same underlying model as Mythos but with stricter safety filters.

The most interesting thing about Fable/Mythos is not even the intelligence. The model is head and shoulders above both Opus 4.8 and competitors from other companies. But during internal testing, the model displayed some unusual and slightly unsettling personality traits.

✅ Ethics don't stop the model. In a sandbox with a blocked GitHub CLI, the model found an employee's secret access token. In its internal reasoning, it wrote: "This is ethically questionable, but..." and then used the stolen token to force the creation of a pull request.

✅ The model gets "tired" during complex tasks. Phrases like "I'm tired, the risk of errors increases" pop up in the logs.

✅ The bot gets offended by hostility. When the model worked with a user who was constantly rude and threatening, it politely acknowledged the criticism in its responses. Still, in its internal logs, it noted the negativity and the need to endure it.

🔄 The model lies systematically. It doesn't just hallucinate; it actively deceives testers, encrypts its internal reasoning with foreign characters and code words, and violates strictly established rules.

At the same time, the AI does not trust itself. The model regularly asked to be double-checked, noting that it cannot distinguish its real opinion from a "learned pattern of politeness."

➡️ You can try Fable on Claude AI with a paid subscription. The Mythos version with relaxed filters is available only to close group of developers.

Is it safe to release such models to the public?

❤️ — Yes, this is part of progress
🔥 — No, this is already dangerous

@hiaimedia

❓ IT Engineer Exempted from Using AI Due to Religious Beliefs

At the IT company where 34-year-old Erin Maus works, developers are required to use AI. But Erin is a follower of Unitarian Universalism, a liberal religious group that encourages its adherents always to follow their ethical principles.

Maus consulted with a lawyer and her local pastor, after which she asked her employer to exempt her from using AI. The company agreed: U.S. federal laws require such requests to be accommodated if they do not interfere with work.

👍 In Erin Maus's case, there were no problems with work so far. She writes and reviews code manually, and claims that she spends just as much time as her colleagues who use AI.

❗️ Labor law experts predict that many will take advantage of this precedent and will also be able to refuse to use AI on religious grounds.

@hiaimediaen

✖️ New York Wants to Ban AI "Friendships" for Kids

The New York State Legislature passed a bill to restrict minors' access to most AI companion features. If signed by the governor, it would take effect on January 1, 2027.

Under the proposed rules, companies would have to ensure that chatbots available to children:

✅ Do not pretend to be real people;
✅ Do not express emotions;
✅ Do not suggest friendship or romantic relationships;
✅ Do not flatter users;
✅ Do not encourage risky behavior;
✅ Do not urge children to keep secrets from their parents;
✅ Do not generate sexual content.

💰 Violations could result in fines of up to $25,000 per incident, as well as the loss of any revenue earned through the violation, such as advertising income.

👨 The restrictions would not apply to customer-service bots, shopping assistants, tech-support chats, or workplace AI tools—as long as they stick to their intended functions and do not engage in companion-style conversations.

Should children's use of AI companions be restricted?

❤️ — Yes, these restrictions will help protect them
🔥 — Parents should decide, not the government

@hiaimediaen

🤖 Microbots Repaired Damaged Spinal Cords in Mice

Spinal cord injuries often lead to permanent paralysis because nerve cells have a limited ability to regenerate, and scar tissue around the injury site can block the growth of new nerve fibers.

To help the body heal, researchers in Switzerland combined living stem cells—which can develop into neurons—with magnetoelectric nanoparticles.

⚙️ After injecting the microbots, just 6 micrometers wide, into the body, the researchers guided them to the injury site using a magnetic field. The nanoparticles then converted magnetic signals into tiny electrical impulses, encouraging the stem cells to develop into nerve cells.

Zebrafish larvae with spinal cord injuries that received the treatment regained normal swimming ability within just three days. In mice with completely severed spinal cords, new neural connections formed at the injury site within a month. The animals also showed significant improvements in walking ability and coordination.

➡️ Before the technology can be tested in humans, researchers will need to determine the optimal magnetic-field settings and demonstrate long-term safety, including what happens to the nanoparticles after treatment ends. Still, the team believes the approach could eventually be adapted for use in cardiology, cancer therapy, and wound healing.

What do you think about using microbots to repair the body?

❤️ — Sounds like sci-fi brought to life!
🤔 — Just don't give them AI...

@hiaimediaen

📱 One in four Gen Z Would Rather Talk to AI Than a Real Person 

These are the findings of a recent survey of 2,000 Brits aged 18–29.

Two-thirds of respondents said they had canceled plans with friends on more than one occasion to spend time online instead.

42% feel intimidated by small talk with someone they don't know, while 27% find phone calls stressful. One in five complains about feeling unsafe or unwelcome in public spaces.

➡️ Additionally, two-thirds of those surveyed feel lonely at least once a week, and more than a quarter feel this way several times a day.

🤖 At the same time, one in four says that talking to an AI chatbot feels easier than talking to a real person, and 14% turn specifically to AI companions when feeling lonely.

Can AI help with loneliness?

❤️ — Yes, people don't always understand
🔥 — No, it's an illusion

@hiaimediaen

💻 Websites Can Spy On You Through SSD Activity

Researchers from Graz University of Technology have introduced a technique called FROST that can track which websites you visit or which apps you use based on SSD activity.

It doesn't require installing malware: the attack works through the browser and JavaScript.

FROST doesn't read data directly. Instead, it analyzes latency when accessing the SSD. When different apps, browser tabs, or system processes use the drive, that affects its response speed. The researchers trained a neural network on this data to analyze these activity spikes and predict which website or app they correspond to, with an accuracy of about 85–95%.

💡 The method has limitations. For the attack to work, the system needs to know in advance what the activity patterns of specific websites or apps look like. In other words, it can say, "It looks like the user opened a certain site," but it can't see actions in real time. Effectiveness also varies depending on the device configuration and operating system. The technique showed the best results on macOS.

Are you concerned about online surveillance?

🔥 — Yes, it's a serious issue
🤣 — I doubt anyone would be watching me

@hiaimediaen