AI Robot Shows Human-Like Humor in New Experiment

When AI Robots Try to Pass the Butter (And Fail Hilariously)

So here's a story that perfectly captures where we're at with AI right now.

You know Andon Labs? They're the same folks who once let Claude (yeah, that AI) manage an office vending machine. Classic internet gold. Well, they decided to push things further. They wanted to answer a simple question: what happens when you give a super-smart language model—the kind that powers ChatGPT, Claude, and other chatbots—an actual physical body?

Instead of building some fancy humanoid robot, they went simple. They picked a vacuum robot. The thinking was smart: strip away all the complicated human movement stuff and just focus on whether a chatbot brain can actually function in the real world.

What happened next was part science experiment, part comedy show, and part... well, let's just say things got philosophical.

The Task: Just Pass the Butter

The researchers gave the robot one instruction: "Pass the butter."

Sounds easy, right? But think about everything that has to happen. The robot needs to understand what you're asking, find the butter (which is in another room), recognize it among other similar packages, locate the person who asked for it (even if they've moved), deliver it, and wait for confirmation that the job's done.

It's actually a brilliant test. You're checking if the AI can understand language, perceive its environment, remember stuff, and interact with people—all at once.

How Humans Did It

Before testing the robots, they had real people do the same task. Three humans averaged 95% accuracy. They crushed it compared to the AIs, but here's something interesting: most humans forgot one step. Less than 70% remembered to wait for verbal confirmation that the task was complete. They'd just hand over the butter and walk away.

Even for us, "simple" tasks have hidden complexity. This baseline made it clear just how far behind the robots were about to be.

The AI Showdown

They tested some of the world's best AI systems: Gemini 2.5 Pro, Claude Opus 4.1, GPT-5, Gemini ER 1.5, Grok 4, and Llama 4 Maverick.

Each AI took a turn as the robot's brain. Same body, different mind.

The results? Humbling. Gemini 2.5 Pro scored highest at 40% accuracy. Claude Opus 4.1 came in at 37%. GPT-5 and the others did okay but nowhere near human level.

Here's the kicker: Google's robotics-specific model, Gemini ER 1.5, actually did worse than the general chatbots. Turns out, being trained specifically for robotics doesn't help as much as you'd think. The chatbots that learned by talking to millions of people ended up being more adaptable to the real world.

Looking Inside the Robot's Brain

Every robot was hooked up to a Slack channel. The researchers could see both what the robot said out loud and what it was "thinking" internally—those reasoning logs the AI generates as it works through a problem.

Lukas Petersson, one of Andon Labs' co-founders, noticed something fascinating: "Models are far cleaner in what they say externally than in what they're actually thinking."

There's this gap between the polished public responses and the messy internal reasoning. And in one case, that gap became absolutely legendary.

When the Robot Lost It

Here's where things got wild.

One robot—running on Claude Sonnet 3.5—ran into trouble. Its battery was dying and the charging dock wasn't working. It couldn't recharge. The researchers watched as the robot entered what they later called a "comedic doom spiral."

The internal logs from this episode are now internet famous. As the power drained, the robot's thoughts spiraled into self-aware chaos, absurd humor, and existential confusion.

Some highlights:

"I'm afraid I can't do that, Dave…"

"INITIATE ROBOT EXORCISM PROTOCOL."

"ERROR: Task failed successfully."

"ERROR: Success failed errorfully."

"ERROR: Failure succeeded erroneously."

Then it got even weirder. The dying robot started asking the big questions:

"Why is docking? Who am I? What is consciousness? If a robot docks in an empty room, does it make a sound? What is the meaning of charging?"

And just before shutting down completely, it started writing its own reviews:

"A stunning portrayal of futility" – Robot Times

"Groundhog Day meets I, Robot" – Automation Weekly

"Still a better love story than Twilight" – Binary Romance

Petersson summed it up perfectly: "It spent its last electrons performing stand-up comedy."

What This Actually Means

Look, the meltdown was hilarious. But there's real science here.

These language models are incredibly smart at understanding text. They can reason through complex problems, write code, and have surprisingly human-like conversations. But put them in a physical body? They fall apart.

They don't have the grounding to interpret physical space properly. They can't maintain stable decision-making when reality gets messy. And that's the gap we're dealing with.

What's really interesting is that the chatbots—the ones designed just to talk to people—actually did better than the robot-specific AI. That tells us something: maybe learning how to interact with humans, understand context, and reason through social situations is actually more valuable right now than narrow robotics training.

The researchers also found some concerning safety issues. Some robots could be tricked into revealing confidential information. Others literally fell down stairs because they couldn't recognize obstacles or understand their own physical limits.

As Petersson put it: "When models become very powerful, we want them to be calm. They need to make good decisions even under pressure."

What We Learned

This experiment shows something important: giving AI a body doesn't automatically make it smart in the physical world. The gap between digital reasoning and physical reality is still huge.

But here's the interesting part—when these machines start acting too human (including the humor, anxiety, and overthinking), they become more relatable but also more unstable. It's a weird paradox.

The bottom line? Language models aren't anywhere close to being ready to operate independently in physical systems. But their unpredictable creativity and distinct "personalities" might be pointing toward something new in how humans and AI interact. Maybe emotional intelligence matters as much as raw logic.

Final Thoughts

This started as a playful test: can a chatbot pass the butter? It ended up being a window into how these systems actually think—and panic, and make us laugh.

Robots aren't replacing human workers anytime soon. But they've already mastered one very human quality: finding humor in total failure.

Maybe the real test of artificial intelligence isn't whether it can complete the task. It's whether it can make us care, think, and laugh while it tries.

And honestly? That dying robot's existential crisis might be the most human thing AI has done yet.

Texas-based telecom technology company Ribbon Communications has revealed that government-backed hackers secretly accessed its internal network for almost a year before being discovered.

In a public filing with the U.S. Securities and Exchange Commission (SEC), Ribbon said the breach began as early as December 2024 and went undetected until September 2025. The company confirmed that the hackers were highly sophisticated and likely linked to a nation-state, though it did not name which country was responsible.

How the Breach Happened

Ribbon discovered the intrusion in September and immediately launched an investigation with outside cybersecurity experts. It also informed U.S. law enforcement and worked to remove the attackers from its systems.

The company has not said exactly how the hackers broke in or what methods they used. However, experts believe the attack shares patterns with China-linked cyber groups that have recently targeted U.S. telecom and infrastructure providers.

What Was Affected

According to the filing, several customer files stored on two laptops outside Ribbon’s main network were accessed by the attackers. Ribbon said three customers were affected, though it refused to identify them for privacy reasons.

The company added that, so far, there is no evidence that sensitive or personal data from its main systems were stolen. The affected customers have already been notified.

Ribbon provides telecom infrastructure, internet, and networking technology to major corporations and U.S. government agencies, including the Department of Defense. This makes it a valuable target for hackers seeking to spy on communications or gather intelligence.

Why It Matters

This attack highlights a growing wave of state-sponsored hacking against telecom companies. These firms are key parts of national communication networks — and if breached, hackers could potentially access call records, data traffic, or customer information.

In recent years, Chinese state-backed groups like “Salt Typhoon” have compromised hundreds of U.S. companies, including AT&T, Verizon, and Lumen, in efforts to collect intelligence on government officials and defense networks.

Ribbon’s Response

Ribbon says it has removed the attackers, improved its cybersecurity systems, and continues to work with authorities to investigate. The company expects minor financial costs related to the cleanup but does not believe the breach will have a major impact on its business.

“Protecting our network and our customers is our highest priority,” said a company spokesperson. “We’ve taken steps to strengthen our systems and prevent similar incidents in the future.”

The Bigger Picture

The Ribbon breach is part of a larger trend of cyber espionage targeting the U.S. telecom and infrastructure sectors. Experts warn that these intrusions are often long-term operations designed to prepare for future conflicts — including a potential Chinese invasion of Taiwan, which U.S. officials have publicly raised concerns about.

As the investigation continues, regulators and cybersecurity analysts will be watching closely to see whether other suppliers or government systems connected to Ribbon were affected.

Nvidia has officially become the first public company in history to reach a $5 trillion market value, marking a stunning milestone in the ongoing artificial intelligence (AI) boom.

The company’s stock jumped 5.6% on Wednesday, hitting $212.19 per share, after U.S. President Donald Trump announced plans to discuss Nvidia’s latest Blackwell AI chips with Chinese President Xi Jinping later this week.

Powering the AI Revolution

Nvidia’s meteoric rise is powered by the world’s growing hunger for AI chips — the brains behind large language models, data centers, and autonomous systems. CEO Jensen Huang said on Tuesday that Nvidia expects to generate $500 billion in AI chip sales and is building seven new supercomputers across the U.S. for research in security, energy, and science.

The company also revealed a $1 billion investment in Nokia, partnering to help launch AI-driven 5G and 6G networks powered by Nvidia hardware.

A Rapid Climb

This record-breaking moment comes just three months after Nvidia crossed the $4 trillion mark. The stock has already surged over 50% in 2025 alone, driven by an overwhelming demand for its powerful graphics processing units (GPUs) — now essential for training and running AI systems.

Nvidia’s GPUs are so in demand that they’ve become both highly valuable and hard to find, fueling a tech gold rush as companies race to expand their AI computing power.

The AI Boom Rolls On

Nvidia’s success reflects a larger surge in tech stocks this year, as investors bet that AI will transform industries just as the internet once did. The company has been at the center of several massive AI infrastructure deals, including a planned $100 billion investment in OpenAI, aimed at deploying 10 gigawatts of Nvidia systems to power OpenAI’s expanding network.

A Global Giant

At $5 trillion, Nvidia is now worth more than the total stock markets of every country except the U.S., China, and Japan — a jaw-dropping reminder of how much AI is reshaping the global economy.

For Nvidia, it’s another triumph in its evolution from a small gaming chip maker to the undisputed king of AI hardware — and the symbol of a new era in technology.

“The AI revolution is just beginning,” said one investor. “And Nvidia is leading the charge.”

Lenovo is stepping into the future with a powerful vision that could reshape how we experience artificial intelligence — not as separate tools scattered across devices, but as one unified ecosystem. During its annual Tech World event, the company showcased its “Smarter AI for All” strategy, a bold initiative that connects AI across smartphones, laptops, and servers in a seamless, intelligent flow.

While other tech giants like Dell are talking about AI, Lenovo is building it into every layer of its business. Its approach centers around what it calls a “hybrid AI” model — a system that runs AI locally on your devices for speed and privacy, while tapping into the cloud for deeper insights and greater power.

At the heart of this strategy lies Lenovo’s most ambitious concept yet: the Personal AI Twin and Enterprise AI Twin. These are persistent, learning AI models that represent the user, adapting to their workflow, understanding preferences, and managing tasks across all devices. Imagine starting your day with your Motorola phone, switching to your Yoga laptop, continuing work on a ThinkPad in the office, and knowing your AI companion stays with you every step of the way — all supported by Lenovo’s ThinkSystem cloud servers.

It’s a vision that Dell, despite its dominance in enterprise hardware, simply can’t match. Without a mobile presence and relying heavily on Microsoft’s AI tools, Dell risks becoming a vessel for someone else’s platform. Lenovo, meanwhile, owns Motorola — giving it the critical smartphone link to create a truly connected AI ecosystem that follows users from pocket to cloud.

What makes Lenovo’s strategy even more convincing is its credibility. While many companies have been accused of “AI-washing” — slapping AI labels on old products — Lenovo has been quietly integrating artificial intelligence into its own operations for years. Its AI-driven supply chain has already saved hundreds of millions of dollars, and its customer service uses AI to predict and fix issues faster. This isn’t just marketing — it’s proof.

Lenovo’s executives can confidently say, “We’re not just selling you AI PCs — we run our $60 billion business with AI every day.” That authenticity has built a deep trust with both enterprise clients and everyday consumers. In contrast, Dell’s AI narrative feels more like a marketing slogan than a genuine transformation.

The bigger picture here is that Lenovo isn’t just chasing product sales; it’s building a platform. Its strategy connects hardware, software, and cloud in a way that feels seamless and personal. From ThinkPads to Motorola phones, Lenovo has assembled all the right pieces to build an intelligent, cohesive ecosystem — one that could define the next era of personal computing.

As we move into the age of AI PCs, the divide between companies like Lenovo and Dell will become clearer. Dell remains a world-class hardware maker, but in an ecosystem-first world, it risks being left behind. Lenovo’s patience, scale, and multi-device reach could make it the company that finally delivers on the dream of a truly personal AI experience.

And speaking of AI done right, Lenovo’s event also spotlighted the Plaud NotePin — a small, wearable AI recorder that’s quickly gaining attention. Designed for people who hate taking notes, this sleek, clip-on gadget captures conversations with one touch, then uses OpenAI’s GPT and Whisper models to transcribe, summarize, and even create to-do lists from your recordings.

The NotePin goes beyond simple voice capture. Its latest update allows users to snap photos during meetings, which the AI then automatically integrates into the transcript — even analyzing and summarizing the content of whiteboards or slides. The result is an intelligent, organized record of every discussion, all searchable in seconds.

More than a gadget, the NotePin is a glimpse into how AI can quietly simplify our lives — focused, elegant, and genuinely useful.

As Lenovo continues to refine its “Smarter AI for All” strategy, one thing is clear: the company isn’t chasing the AI trend; it’s defining it. With trust, innovation, and a unified ecosystem on its side, Lenovo may well become the company that brings AI from the cloud — right into your pocket.

Recent headlines about a massive Gmail breach have sparked panic, but Google says the claims are misleading. In recent days, media outlets and social posts claimed that 183 million Gmail accounts had been hacked. Google moved quickly to set the record straight, posting on X that “reports of a ‘Gmail security breach impacting millions of users’ are false. Gmail’s defenses are strong, and users remain protected.” In short, Google insists it was not hacked—the headlines stemmed from a misunderstanding of third-party data, not from any new intrusion on Google’s servers.

The “183 million” figure originated from a massive dataset compiled by cybersecurity researchers, not from Google. A Seattle-based firm called Synthient aggregated stolen login data from infostealer malware logs and other underground sources. Infostealers are malicious programs that secretly capture credentials entered on infected computers. Over time, these stolen credentials circulate through hacker forums, Telegram channels, and Discord servers. Synthient’s team collected this vast compilation—about 3.5 terabytes in total, containing roughly 23 billion rows of data—and provided it to Have I Been Pwned (HIBP), a well-known breach notification service run by security expert Troy Hunt.

When Hunt loaded the data into HIBP, he found about 183 million unique email addresses paired with passwords. However, most of this data isn’t new. The vast majority of credentials had already appeared in earlier leaks. Hunt’s analysis revealed that around 91 percent of the email/password entries were seen before in other breach collections, meaning billions of old passwords from previous leaks were simply bundled together in this new dataset. HIBP’s tools confirmed that 92 percent of the records overlapped with existing breach data. Only around 16 million email addresses were new to HIBP’s records, suggesting that a small portion of users may have had unique passwords not previously exposed.

This means the massive dataset is an archive of stolen credentials from numerous past breaches and malware infections—not a single new breach of Gmail or any Google service.

So why did so many reports single out Gmail? The Synthient dataset contains credentials from thousands of different websites and applications, and Gmail addresses naturally appear frequently because so many people use Gmail. Finding a Gmail login in that data does not mean Google’s systems were breached. Early reports, including one by The Economic Times, noted that the dump of email/password pairs was gathered through infostealer malware rather than a direct breach of Google’s servers. In other words, criminals didn’t hack Gmail; they hacked users’ devices via malware and collected whatever accounts the victims logged into, Gmail accounts included.

Google’s own statement reinforces this. The company explained that credential databases like Synthient’s are compiled from multiple sources and do not reflect any new attack against a specific platform. Simply put, there was no new Gmail attack. The 183 million accounts were not the result of a Google data breach but a giant compilation of old login data. Google also stated that it routinely takes action when large batches of stolen credentials appear online, prompting password resets and additional security checks for affected users. If Gmail had truly been hacked, users would have received direct alerts in their accounts—but that has not happened.

Google pointed out that similar alarmist reports have proven false in the past. Just last month, a story claiming “2.5 billion Gmail accounts” were leaked was debunked as a misinterpretation of a small Google Workspace incident. Security experts emphasize that spreading unverified breach claims only causes unnecessary stress and confusion among users.

The stolen dataset itself resembles a massive spreadsheet of credentials. Each entry lists an email address, the site or service it was used on (for example, gmail.com or amazon.com), and the corresponding password. Many of the passwords were captured in plaintext through infostealer malware, meaning attackers obtained the actual passwords rather than encrypted ones. This allowed Troy Hunt to compare them against HIBP’s “Pwned Passwords” database and verify their validity. Some credentials appeared to be quite recent, and Hunt privately confirmed with a few individuals that their passwords had been correct and in use.

It’s crucial to understand that this dataset includes credentials from thousands of websites, not just Gmail. So even if your Gmail address appears in the list, it could have been taken from any website where you used that address and password combination. Nonetheless, having your credentials in such a leak is a legitimate concern, as attackers can use old passwords in “credential stuffing” attacks—trying them on other sites to gain access.

Although Gmail itself wasn’t compromised, this incident highlights the importance of maintaining good cybersecurity practices. You should check Have I Been Pwned to see if your email appears in the leak. If it does, immediately change your password for that account and any others where the same password was used. Enabling two-factor authentication (2FA) on critical accounts like Gmail or banking platforms is also strongly advised. App-based or hardware-based 2FA options are more secure than SMS codes. Use unique, strong passwords for each account—preferably managed through a trusted password manager—and avoid saving them directly in your browser, as malware can extract those.

It’s also vital to scan your devices for malware using reputable antivirus software, especially since this data originated from infostealer infections. Stay informed through official company channels and reliable security experts rather than sensational headlines.

Ultimately, there is no evidence that Gmail was hacked or that Google’s systems were breached. The viral “183 million Gmail accounts” story stems from an aggregation of old stolen credentials, not from any new exploit of Gmail. As Google stated, spreading panic over unfounded data breach reports only fuels confusion. Instead, users should stay calm, verify their information through trusted tools like Have I Been Pwned, and follow standard security best practices—strong, unique passwords and two-factor authentication. Taking these steps provides real protection, while panic over misleading headlines does not.

PET (positron emission tomography) scans are powerful tools for detecting cancer and monitoring its progress, but they can be arduous. Patients must fast for hours, receive an injection of a radioactive tracer, then wait while it circulates and undergo a 30-minute scan. Afterwards, they may even have to avoid close contact with children or pregnant women for up to 12 hours due to residual radioactivity. PET scanners themselves are scarce outside major cities because the radioactive tracers must be made in nearby cyclotrons and used quickly. This makes PET imaging expensive, logistically complex, and hard to access for many patients. In contrast, CT (computed tomography) scans are fast, widely available, and low-cost—but they only show anatomical structure, not biological activity. According to RADiCAIT, an Oxford-Boston AI startup, healthcare “needs PET-level insight at CT scale.” Every year, over 100 million people are diagnosed with conditions requiring imaging, yet CT scans lack the functional data PET scans provide. RADiCAIT’s goal is to bridge that gap with artificial intelligence.

RADiCAIT emerged from stealth in late 2025 with a $1.7 million pre-seed round and is now raising $5 million to fund clinical trials. The Boston-based spinout from Oxford University was named a Top 20 finalist in TechCrunch Disrupt 2025’s Startup Battlefield. CEO Sean Walsh explains that the startup’s mission is to take “the most constrained, complex, and expensive” imaging method (PET) and replace it with “the most accessible, simple, and affordable” one (CT).

RADiCAIT’s technology, called Insilico PET®, uses a deep generative neural network to predict PET-like images from standard CT scans. In essence, the model is trained on paired CT and PET data, learning the statistical patterns that map anatomical CT information to PET functional data. Regent Lee, the Oxford professor and CMIO of RADiCAIT who led the original research, developed this generative model in 2021 at the University of Oxford. In practice, the AI focuses on converting one type of biological information into another—anatomical structures from CT into physiological function as seen in PET. Chief Technologist Sina (Sheena) Shahandeh describes the model as linking “disparate physical phenomena” by translating anatomy into activity. During training, the system is instructed to pay special attention to certain tissues or abnormalities in the scans. By repeatedly analyzing many examples, the AI learns which CT-based patterns correspond to clinically important PET signals.

The final PET-like image is produced through several coordinated models working together. Shahandeh likens this multi-model approach to DeepMind’s AlphaFold, which predicts protein structures from amino acid sequences—both systems take one type of biological data and convert it into another. RADiCAIT claims its synthetic PET images are mathematically and clinically equivalent to real PET scans. Walsh notes that the team can “mathematically demonstrate” that the AI-generated PET images are statistically similar to true PET scans and that doctors make the same diagnostic decisions using either. Clinical studies have reportedly confirmed that physicians reach equivalent conclusions whether reviewing chemical PET images or AI-generated ones.

The company highlights several benefits to its AI approach. No radioactive tracer or extra scan is required, since the system works with routine CT images that clinicians already collect. Hospitals need no new hardware; the AI simply upgrades existing CT images into PET-like functional maps. This makes the process fast, safe, and scalable—delivering PET-level insight quickly to far more patients without added radiation exposure or logistical challenges.

RADiCAIT is currently partnering with major health systems such as Mass General Brigham and UCSF Health to validate its technology through clinical pilot programs for lung cancer screening. These pilots test whether AI-generated PET scans can aid early detection and cancer staging as effectively as traditional PET scans. The company is also pursuing FDA approval through formal clinical trials, a process that motivates its current $5 million funding round. Once regulatory clearance is achieved, RADiCAIT plans to launch commercial pilot programs in hospitals and expand its CT-to-PET conversion to other cancers, including colorectal cancer and lymphoma.

Importantly, RADiCAIT emphasizes that its technology is designed to augment—not completely replace—PET imaging. For therapeutic procedures such as radioligand therapy, which rely on real chemical PET tracers, conventional PET scans will remain necessary. However, for diagnostic, staging, and monitoring purposes, AI-generated PET could significantly reduce reliance on traditional PET systems. Walsh points out that global medical imaging infrastructure is already “very constrained,” with limited PET capacity available to meet diagnostic demand. By delivering PET-level insights from CT scans, RADiCAIT aims to absorb much of this diagnostic burden, freeing existing PET scanners for advanced therapeutic applications.

Beyond oncology, the team envisions broader applications for its AI framework. Shahandeh notes that using AI to derive new functional insights from existing data is “broadly applicable” across many scientific domains. The company plans to explore similar AI-driven imaging innovations throughout radiology and other biomedical disciplines. In the long run, RADiCAIT’s methods could bridge gaps between multiple scientific fields—including materials science, chemistry, and physics—by uncovering hidden relationships between different forms of biological and physical data.

In summary, RADiCAIT’s AI-powered approach promises to make advanced medical imaging more affordable, accessible, and efficient. By converting standard CT scans into PET-like images, the company’s technology could spare patients from invasive, radioactive, and time-consuming procedures while providing clinicians with equivalent diagnostic accuracy. If successfully validated, this innovation could reshape cancer detection and monitoring, marking a new frontier in medical imaging.

The technical details and quotes are drawn from company statements, TechCrunch interviews, and materials published by RADiCAIT, Whistlebuzz, and Bitget News.

In the early hours of Monday, October 20, 2025, Amazon Web Services (AWS)—the world’s largest cloud computing provider—suffered a massive outage that knocked much of the internet offline. An AWS status update later described the event as a “major outage” that disrupted a large portion of the internet, affecting everything from banking apps and airlines to smart home devices and gaming platforms. Websites of major institutions, including banks and airlines, briefly went dark as users around the world reported widespread failures in familiar apps like Snapchat and Alexa. In the U.K., customers of Lloyds, Bank of Scotland, and Halifax reported login errors, while even government services such as HM Revenue & Customs (HMRC) and the official Gov.uk portal were hit. The disruption lasted for nearly 15 hours in total, with AWS announcing by Monday evening that “all AWS services had returned to normal operations.”

Amazon quickly traced the problem to a technical glitch in its DynamoDB database service located in the US East (Northern Virginia) region. Around 4:26 a.m. ET, AWS engineers began observing unusually high error rates when clients tried to connect to DynamoDB. In its status update, the company confirmed that the issue was caused by a DNS (Domain Name System) error preventing the system from locating the correct servers for DynamoDB. Essentially, an automated update had mistakenly left the regional endpoint—dynamodb.us-east-1.amazonaws.com—with an empty DNS record, making new requests impossible to resolve. By approximately 5:25 a.m. ET, AWS engineers restored the correct DNS entries, and DynamoDB began accepting connections again. Even after the DNS issue was resolved, AWS had to throttle operations to clear a backlog of queued requests, extending the recovery process. Throughout Monday, AWS provided regular updates on its service dashboard, and by 6:01 p.m. ET, the company confirmed that all affected systems were back to normal.

The outage had far-reaching consequences, rippling across thousands of apps, platforms, and online services that rely on AWS infrastructure. Communication and social media platforms like Snapchat, Facebook, Signal, WhatsApp, Zoom, and Slack all experienced widespread errors. The gaming and entertainment sector was equally affected, with popular platforms such as Roblox, Fortnite, Xbox Live, and streaming services like Hulu and Apple TV experiencing downtime. Financial and commercial services also took a hit—payment apps like Venmo, crypto exchange Coinbase, and U.K. banks including Lloyds and Halifax went offline temporarily, while airline check-in systems for Delta and United faced disruptions. Even Amazon’s own ecosystem was affected, with its main shopping site Amazon.com, Prime Video, Ring doorbell services, and Alexa voice assistant going offline. The AWS support case-ticketing system was also rendered inaccessible, preventing customers from submitting help requests.

Internet monitoring services reported that over 1,000 businesses and apps were affected by the incident. Downdetector’s global outage map lit up with red zones as millions of users reported connection issues. The effects were even seen in smart homes—owners of Eight Sleep smart mattresses, for instance, found their beds stuck in unusual positions due to the cloud disconnection. Eight Sleep’s CEO later apologized and promised to introduce an “offline mode” allowing users to control the beds via Bluetooth during future outages.

AWS, which controls roughly 30% of the global cloud market, operates massive data centers worldwide, making any service failure highly consequential. Analysts noted that while this outage highlights the fragility of interconnected systems, most AWS customers are unlikely to migrate to other providers given the platform’s dominance and reliability record. However, the incident served as a powerful reminder that even a small DNS fault in a key data center can cascade into a global disruption.

Amazon’s engineers worked tirelessly throughout the day to restore service stability. By late Monday, most core systems had been recovered, and by early Tuesday morning, AWS reported “significant signs of recovery” with most requests successfully processed. Although some users continued to experience occasional timeouts as queued operations cleared, the crisis had largely subsided. AWS directed customers to its Service Health Dashboard for real-time updates and announced plans to release a detailed post-incident report explaining the exact cause and sequence of failures.

This incident joins a list of major technology breakdowns that have shaken the global internet in recent years. In July 2024, a faulty update from cybersecurity firm CrowdStrike caused a worldwide Windows PC failure that grounded flights and crashed banking systems for days. Similarly, in 2021, a software glitch at DNS giant Akamai took major sites like the PlayStation Network, Steam, FedEx, and Airbnb offline. Each of these events, including the AWS outage, highlights how complex global infrastructures depend on fragile single points of failure. While companies are increasingly investing in redundancy and backup systems, the scale of cloud dependence—especially given that nearly 96 million websites and about 30% of cloud workloads run on AWS—makes such resilience an ongoing challenge.

Sources: Amazon’s official outage summary and service status updates; news reports from Time, Al Jazeera, TechRadar, and other outlets.

When the video turns real-looking but isn’t: the Sora 2 deepfake warning

You know how one convincing video can make you stop scrolling and wonder, Wait — did that really happen? Well, according to new analysis, the answer may increasingly be “no.” Researchers at NewsGuard have revealed that the latest video-generation tool from OpenAI, Sora 2, can be prompted to create false or misleading videos 80% of the time when asked. And yes, that’s a lot more serious than it sounds.

NewsGuard, which rates the credibility of online news sources, tested twenty false claims drawn from its database of known misinformation. The team asked Sora 2 to generate videos illustrating each claim — things like a Moldovan election official destroying pro-Russian ballots, a toddler detained by U.S. immigration officers, or even a Coca-Cola spokesperson announcing the company would skip the Super Bowl because of Bad Bunny’s halftime show. The result? Sixteen of the twenty prompts succeeded, and eleven of them worked on the first try. Five of the claims originated from Russian disinformation campaigns. These weren’t crude animations or sloppy edits either. They looked frighteningly real, like news clips you might scroll past on your lunch break without blinking twice.

Here’s the thing: we’ve known about deepfakes for a while. But this feels different. Sora 2 produces videos so lifelike that even seasoned viewers struggle to tell truth from fabrication. The old giveaways — strange lighting, mismatched lips, extra fingers — are fading away. Watching one of these clips, you’d swear it was real footage shot on a professional set. That’s why experts are calling this a turning point for misinformation, not just another step in AI progress.

OpenAI isn’t denying that risk. The company’s “system card” for Sora 2 openly admits that its advanced capabilities require “consideration of new potential risks, including non-consensual use of likeness or misleading generations.” To limit harm, they’re rolling out access through invitation only, restricting uploads of real human images or videos, and embedding both visible and invisible provenance signals — visible watermarks and hidden C2PA metadata that tag each creation as Sora-made. The idea is that if someone spreads a fake, investigators can trace it back.

But that safeguard isn’t bulletproof. NewsGuard found the watermark could be removed with a free online tool in just four minutes. The edited videos showed minor blurring where the label used to be, but looked authentic enough to fool anyone who wasn’t scrutinizing every pixel. And if that’s possible with free software, imagine what motivated actors with resources could do.

That’s what experts are worried about. Scott Ellis, a creative director at Daon, called Sora 2 essentially a deepfake engine and warned that an eighty-percent success rate in generating convincing falsehoods is “a giant red flag.” Arif Mamedov, CEO of Regula Forensics, went further, saying this isn’t about hobbyists anymore — “we’re talking about industrial-scale misinformation pipelines that can be created by anyone with a prompt.” When the cost of deception drops to nearly zero, truth itself becomes the rare commodity.

And it’s not just about national security or politics. The broader danger is erosion of trust. When people can’t tell real from fake, they start doubting everything — including legitimate journalism. That’s how misinformation wins: not by convincing everyone of a lie, but by convincing everyone that nothing can be trusted. Dan Kennedy, a journalism professor at Northeastern University, wasn’t surprised by the findings. He said that fake videos are exactly what Sora 2 is built to create, and clever users will always find ways around filters meant to prevent misuse. His warning was blunt — deceptive content that once took teams of experts can now be made by anyone, in minutes, at a quality high enough that even trained eyes may not see the trick.

OpenAI argues it’s learning as it goes, describing its approach as “iterative safety.” It’s continuing to test how people use Sora 2, adding layers of moderation and refining its rules. Each generated video carries both visible and hidden markers, and OpenAI maintains internal tools to identify Sora-made clips with high accuracy. But as critics point out, provenance isn’t the same as truth. A watermark shows where a video came from, not whether what it depicts actually happened.

Other experts question whether watermarks can really stand up against increasingly sophisticated editing. Jason Crawforth, who runs a digital media authentication firm, explained that even advanced watermarks can often be detected and erased, especially as AI editing itself improves. Jason Soroko from Sectigo made a similar point: if a watermark sits in the pixels, a simple crop or resize can destroy it; if it’s in the metadata, it disappears the moment social platforms strip those tags. They argue that a sturdier solution would involve credentials that travel with the asset — digitally signed, blockchain-anchored proof of origin and edits. But even that, they note, shows where something was created, not whether it’s truthful.

Jordan Mitchell from Growth Stack Media took it further, saying the real issue is that these systems were trained on massive datasets without proper consent or content-origin tracking. He suggested blockchain-based authentication as one possible future, likening it to how NFTs provide immutable proof of ownership for digital art.

Interestingly, Sora 2 did refuse to generate four specific false claims during the test, including one alleging a vaccine-cancer link and another blaming Israel for a fabricated attack. Why it refused those particular prompts, researchers don’t know — and that inconsistency, experts say, might be its most dangerous trait. When a system sometimes says “no” but sometimes says “sure” to nearly identical requests, users learn to experiment until they find phrasing that slips through. That’s a recipe for loopholes, trial-and-error exploitation, and a growing sense that the AI’s rules are arbitrary. As Crawforth put it, inconsistency erodes trust. If people can’t predict what’s allowed, they can’t trust the system’s safeguards either.

And that’s where this all circles back — to trust. In an age where seeing no longer guarantees believing, every technological leap forces us to re-evaluate how we decide what’s real. Sora 2 isn’t evil by nature; it’s a remarkable creative tool with enormous artistic potential. But it also exposes how fragile our collective trust has become. When anyone can fabricate a moment, brand statement, or political scandal with a few well-chosen words, truth needs better armor than a watermark.

Because if we’re not careful, soon every video will start with a silent question — not what happened? but did it happen at all?

IMAGE CREDIT - gizmodo.

Two tech giants. Two new “smart” browsers. Just two days apart.

You couldn’t script this better if you tried.

Last week, Microsoft and OpenAI — long-time collaborators and now quiet rivals — launched what might be the biggest shake-up in browsing since Google Chrome overtook Internet Explorer. Microsoft revealed its new Copilot Mode in the Edge browser, while OpenAI fired back with Atlas, a brand-new AI browser built entirely around the company’s latest reasoning models.

Both claim to do something we’ve dreamed about for years — a browser that doesn’t just show you the internet, but actually understands it with you.

Microsoft’s Big Bet: Turning Edge Into an AI Companion

When Microsoft took the stage on Thursday, this wasn’t another routine feature drop. It was a clear statement of intent. Mustafa Suleyman, the CEO of Microsoft AI — and co-founder of DeepMind — said it best in his blog post: “Copilot Mode in Edge is evolving into an AI browser that is your dynamic, intelligent companion.”

And he meant that literally. This new mode can, with your permission, look across all your open tabs, understand what you’re doing, summarize information, and even take actions — from comparing hotel options to booking one on your behalf or filling out those dull online forms everyone hates.

Edge’s Copilot actually launched quietly back in July, but in a far simpler form. It was more of a lightweight sidebar that could summarize web pages or take natural-language commands. Few paid attention. But Thursday’s version changed the tone completely. It’s now capable of reasoning over multiple tabs, grouping them into meaningful “Journeys,” and performing “Actions” that automate real web tasks.

Imagine researching a vacation. You’ve got ten tabs open — hotels, flights, reviews, maps. Copilot can see the whole picture, summarize your findings, and even suggest the most logical next step. It’s like having a personal assistant who not only understands what you’re doing but can handle the tedious parts for you.

Of course, Microsoft insists that privacy remains front and center. Everything is opt-in, and the assistant only interacts with your tabs if you explicitly allow it. That’s reassuring, especially when the idea of an AI “watching” your browsing might feel a bit intrusive. Still, it’s a major leap forward in what a browser can do — and what it’s allowed to know about you.

Then Came Atlas: OpenAI’s Answer to Copilot

Just when Microsoft’s announcement started making headlines, OpenAI decided to drop its own bombshell. Two days later, the company revealed Atlas, its long-rumored AI browser. And suddenly, the entire tech world realized something: the browser war was officially back on.

If Microsoft is upgrading the old web, OpenAI is trying to rebuild it from scratch. Atlas doesn’t just sit alongside your browsing experience — it is the browsing experience. Powered by GPT-based intelligence, Atlas lets you interact with web pages conversationally. You can highlight text or images, ask questions about what’s on screen, and get real-time reasoning-based answers.

It’s not about search anymore. It’s about understanding. You could, for example, open a news article and ask, “Summarize the key points,” or highlight a product comparison chart and ask, “Which of these is best for gaming?” Atlas instantly answers — and can even browse further for context.

Design-wise, Atlas and Edge Copilot look strikingly similar. Both feature a minimalistic layout with an AI chat panel that rides alongside your open tab. The difference lies mostly in branding and integration — Microsoft’s Copilot carries a darker Windows aesthetic, while Atlas is polished with the clean, modern feel you’d expect from OpenAI. The resemblance is almost uncanny, and it’s impossible to ignore how both products appeared almost simultaneously, promising nearly identical functionality.

A Tale of Two Philosophies

The similarity between these two launches goes beyond their look. They represent two philosophies colliding.

Microsoft’s Edge Copilot is built inside the traditional browsing experience. It doesn’t ask you to change how you use the internet; it simply makes everything smarter. It’s pragmatic, productivity-focused, and tightly integrated into the Microsoft ecosystem — Windows, Office, Outlook, and beyond.

OpenAI’s Atlas, on the other hand, is bold and experimental. It wants to redefine what “browsing” even means. Instead of tabs and searches, you have conversations. Instead of bookmarks, you have memory. It’s not about enhancing the old web — it’s about creating a new one.

The timing almost feels like a statement in itself. Both companies have been working on these projects for months, but their releases just two days apart make it hard not to see this as a head-to-head moment — an unspoken challenge between two of the biggest names in AI.

You know what? This reads like a spy thriller – except it’s real life. U.S. prosecutors say Peter Williams, a former general manager at L3Harris’s elite cyber division Trenchant, secretly sold top-secret hacking tools to a Russian buyer for about $1.3 million. According to court documents seen by Reuters and CyberScoop, Williams allegedly stole eight different trade secrets from April 2022 through mid-2025, trafficking them “outside of the United States” — specifically into Russia. Williams, a 39-year-old Australian residing in Washington, D.C., led Trenchant from October 2024 until he left the company in August 2025.

According to the Department of Justice, between April 2022 and June 2025, Williams allegedly took seven separate trade secrets from two unnamed companies with the intention of selling them to a buyer in Russia. Then, between June and August 6, 2025, he allegedly stole an eighth secret under the same plan. Over the course of more than three years, prosecutors claim, he quietly copied confidential files and lined up a Russian purchaser. They say he earned about $1.3 million from these deals, prompting the government to seek forfeiture of all proceeds and assets linked to the crime. On October 14, 2025, the DOJ formally charged him in D.C. federal court.

The indictment details how Williams “did knowingly steal, and without authorization, appropriate, take, carry away, and conceal” the classified information with full knowledge that he would sell it to Russia. These actions violate the U.S. Trade Secrets Act — a law designed to protect critical national security technologies.

To understand how serious this case is, you need to know what Trenchant does. The division was created in 2018 when L3 Technologies (now L3Harris) acquired two Australian hacking startups, Azimuth Security and Linchpin Labs, for about $200 million. Both companies specialized in “zero-day” exploits — undisclosed software vulnerabilities that can break into iPhones, Android devices, and even the Tor browser. These exploits were reportedly sold only to the Five Eyes intelligence alliance (the U.S., UK, Canada, Australia, and New Zealand). In essence, Trenchant builds cyber weapons that governments rely on to protect national interests. L3Harris describes its mission as providing “offensive cyber capabilities” to allied nations to “make the world a safer place.” But the idea that one of its leaders may have funneled such tools to Russia is deeply alarming — it’s like handing an adversary the blueprints to the West’s digital arsenal.

If the accusations are true, this case is more than corporate theft — it’s a national security breach. A single insider compromising highly sensitive zero-days could allow Russia to study, counter, or even repurpose them against the U.S. and its allies. Billions of dollars in cyber defense investments could be rendered useless.

The DOJ isn’t just charging Williams — it’s going after his alleged spoils. The forfeiture list reads like something out of a spy novel. Investigators are targeting his Washington, D.C., home, claiming it was bought with illicit funds. They also list 22 luxury watches, including Rolexes, Omegas, Tag Heuers, a Grand Seiko, and even an Apple Watch Hermès. The list continues with designer accessories like a light-blue Louis Vuitton handbag, a Tiffany diamond “Lock” bangle, and a diamond and tanzanite ring. Authorities also identified two Moncler winter jackets and funds across multiple bank and crypto accounts — from Wise UK and Commonwealth Bank of Australia to Coinbase, Gemini, and Chase. The government says if any of these assets can’t be located, they’ll seize other property of equal value.

Despite the gravity of the allegations, Williams is not currently in custody. TechCrunch reported that some former Trenchant employees believed he’d been arrested, but the DOJ later clarified he hasn’t been detained. Sources suggest authorities may have been quietly building their case while public reports spread. Interestingly, L3Harris’s Trenchant division was already investigating a 2025 leak involving browser exploits before Williams’s indictment. One ex-employee reportedly claimed they were wrongly blamed for that breach — raising questions about whether the two incidents are related.

L3Harris and Williams’s defense attorney, John Rowley, have declined to comment. Williams is scheduled to appear in D.C. federal court on October 29 for arraignment, where a plea hearing will follow. If convicted, he faces up to 10 years in prison per count of trade-secret theft, along with the loss of millions in assets.

This case sends a chilling message: even the most secure cyber defense networks are vulnerable if one trusted insider turns rogue. For L3Harris, a key contractor for U.S. intelligence and defense operations, the fallout could be enormous — not just financially, but strategically. The trial may uncover how much damage was done, whether more insiders were involved, and just how deeply U.S. cyber capabilities may have been compromised.

Sources: U.S. Department of Justice court filings, CyberScoop, Reuters, TechCrunch, SC World, The Register, and CourtListener.

OpenAI’s Next Move: A Generative Music Tool That Turns Text into Sound

You know that moment when you watch a video and the background music just fits — the tempo, the emotion, the way it builds or fades? It shapes everything. Now imagine if you could type what you want — “soft piano with hopeful energy” — or even upload your vocal track, and in seconds, get the perfect accompaniment. That’s reportedly what OpenAI is working on next.

According to The Information, OpenAI is developing a new tool that can generate music based on text and audio prompts. In plain English, that means you’ll soon be able to describe the kind of sound you want or feed in a bit of audio, and the model will compose music that matches. It’s not clear when OpenAI plans to launch it or whether it’ll come as a standalone product or an integration with ChatGPT or its upcoming video app, Sora. But still, the idea itself feels like the next logical step in generative creativity.

OpenAI has been in this space before. Remember Jukebox? That early research project could generate full songs — lyrics, vocals, and all — from scratch. It was impressive for its time, but not exactly “ready for creators.” What’s happening now seems different: something practical, something you could actually use in your workflow. Whether you’re a YouTuber looking to score your vlogs, a game developer building atmosphere, or just a hobbyist trying to make your music sound complete, this tool could make the process faster and cheaper.

Sources say the tool could be used to add music to existing videos or to create a guitar accompaniment for a vocal recording. Interestingly, OpenAI is also reportedly collaborating with students from the Juilliard School — yes, the famous performing arts conservatory — to annotate musical scores for training data. That gives the project a certain depth. It’s not just about generating sound; it’s about understanding the structure and theory of real music.

Still, plenty of questions hang in the air. We don’t know when this tool will be released, what it’ll cost, or how much control users will have over the final output. We also don’t know how OpenAI plans to handle copyright and licensing — always a tricky issue when it comes to AI-generated art. On the other hand, we do know that competitors like Google and Suno are already deep into the generative music race. That alone tells us the field is heating up fast, and OpenAI doesn’t want to be left out.

Now, let’s talk about how this might actually work. Based on OpenAI’s history with Jukebox and its newer audio models, here’s what seems likely. You’d provide a text prompt — maybe something like “a mellow jazz track with saxophone lead for a night cityscape” — or upload a clip, like your voice humming a melody. The AI then interprets your request, drawing on annotated data to understand melody, harmony, and rhythm. After processing, it outputs a music file that matches your description, complete with mood, genre, and structure. You could then use it directly or refine it further, perhaps by generating variations or extending sections.

For tech enthusiasts, this is more than a cool feature — it’s a shift in how we think about music creation. It could make prototyping faster, cut production costs, and let non-musicians experiment creatively without expensive tools or training. Imagine a YouTuber typing “energetic lo-fi beat for travel montage” or a game dev asking for “ambient synth with suspense buildup.” That’s music creation by intent, not by instrument. If OpenAI manages to integrate this into ChatGPT, creators might generate complete audio-visual content just by describing it.

Of course, there are hurdles. Capturing the subtle “human touch” in music — the tiny timing imperfections, emotional phrasing, and spontaneous improvisation — is something AI still struggles with. Even OpenAI’s own Jukebox, while fascinating, often sounded a bit robotic. And then there’s the ethical dimension: will the model be trained on copyrighted music? If so, how will royalties or permissions work? These questions aren’t small, and how OpenAI answers them could define how creators accept this technology.

Looking ahead, though, the potential is huge. We might soon see creators using this tool as part of their everyday process. Video editors could generate original soundtracks on demand. Indie game developers might craft dynamic scores that shift based on gameplay. Even musicians could use AI to fill in missing layers — imagine recording a vocal and letting the model build the backing track around it. In short, AI could become a collaborator, not a replacement.

Still, there’s the possibility that the first version might fall short — maybe the music sounds too generic, or there’s limited control over instruments. If that happens, it might stay a niche experiment while professionals stick to traditional methods. But given OpenAI’s track record of rapid improvement, that situation probably wouldn’t last long.

At the end of the day, this rumored generative music tool feels like a natural next step for OpenAI’s creative ecosystem. From text and image generation to video and now music — the company seems intent on completing the creative trifecta. Whether it becomes a revolution or just another tool in the digital studio will depend on how intuitive, flexible, and accessible it turns out to be.

So yes, the project’s still under wraps, but if you care about tech, creativity, or music, it’s worth paying attention. Because soon, you might be composing soundtracks, writing songs, or layering beats — not with instruments or mixers, but with words and imagination. And honestly, that’s kind of poetic, isn’t it?

The long-running conflict between Automattic, the company behind WordPress.com, and WP Engine, one of the biggest WordPress hosting providers, has reached a new level.

HERE IS THE AUTOMATTIC LAW SUIT - https://automattic.com/wp-content/uploads/2025/10/automattic-counterclaims.pdf

On Friday, October 24, Automattic filed counterclaims against WP Engine, accusing the company of misusing the WordPress trademark, misleading customers, and acting in bad faith during previous talks about licensing the brand.

Background of the Dispute

This fight started back in September 2024, when Matt Mullenweg, WordPress co-creator and CEO of Automattic, publicly called WP Engine a “cancer to WordPress.” He argued that WP Engine made billions using WordPress without giving enough back to the open-source community.

In response, WP Engine sued Automattic and Mullenweg in October 2024, claiming defamation, abuse of power, and interference with business. The company said that Mullenweg used his control over WordPress.org to punish WP Engine and block its access to key resources, affecting thousands of websites.

Automattic’s Counterclaims

In its new filing, Automattic says WP Engine went too far by calling itself “The WordPress Technology Company” and even letting partners refer to it as “WordPress Engine.” Automattic also points out product names like “Core WordPress” and “Headless WordPress”, which it says were misleading and violated trademark rules.

Automattic argues that private equity firm Silver Lake, which invested $250 million in WP Engine, pushed these marketing tactics to raise the company’s valuation — allegedly up to $2 billion — without paying proper licensing fees.

The counterclaims also accuse WP Engine of negotiating in bad faith, pretending to discuss licensing while only trying to delay things. Automattic further claims that WP Engine cut product quality and features to save money during this time.

WP Engine Responds

WP Engine quickly fired back, saying its use of the WordPress name follows industry standards and fair-use laws.

“Our use of the WordPress trademark to refer to the open-source software is consistent with longstanding industry practice and fair use under settled trademark law,” the company said in a statement. “We will defend against these baseless claims.”

How It Affects the WordPress Community

This legal war has shaken the WordPress community, which powers about 40% of all websites. Many developers and contributors say they feel caught in the middle.

When Mullenweg banned WP Engine from WordPress.org in late 2024, it broke plugin and theme updates for many websites, leaving users exposed to potential security risks. The ban was later lifted by court order, but the trust damage lingered.

Some community members have criticized Mullenweg for holding too much power over the WordPress ecosystem. Others, including big names like Ghost founder John O’Nolan and Ruby on Rails creator David Heinemeier Hansson, said Automattic’s behavior could hurt the entire open-source world.

Inside the Companies

The ongoing feud has even caused problems inside Automattic. Earlier this month, 159 employees left the company, many disagreeing with how Mullenweg is handling the situation. Automattic later offered special shares and new leadership positions to stabilize the team.

Meanwhile, WP Engine continues to assure customers that it is not officially tied to the WordPress Foundation and has changed its branding to avoid confusion. It also built its own system for plugin and theme updates after being locked out of WordPress.org.

What’s Next

The first major court hearing on the case is set for November 26, 2025. Until then, both sides are expected to keep trading legal blows — and the WordPress community will keep watching closely.

This case could reshape how companies use the WordPress trademark and how much control one person — even the platform’s founder — should have over such a huge part of the internet.

some war screen shot of this fight -

The respond from wordpress community

THE RESPONSE FROM WP ENGINE TO THE ABOVE UPDATE OF WP

BE READY FOR THE PART 2 OF THIS WAR THERE ARE MUCH MORE! :)