TikTok is stepping up its direct messaging game with new features like voice notes and the ability to share up to nine photos or videos in chats. The update makes TikTok’s messaging feel more like WhatsApp or Instagram, while giving Gen Z a new way to connect beyond just sharing videos.
Key Takeaway: With voice notes up to 60 seconds and safer media sharing options, TikTok aims to be more than just an entertainment app — it wants to become a regular communication hub for its users.
Cybersecurity experts have uncovered a new AI-powered ransomware called PromptLock, and it’s making waves in the security world. Unlike traditional ransomware, PromptLock uses advanced AI techniques to execute attacks more efficiently, making it harder to detect and defend against.
What makes PromptLock even scarier is its ability to adapt and change its attack patterns, which means standard security tools might not be enough. Experts warn businesses and individuals to update their systems, use strong authentication, and keep regular backups to stay safe.
This discovery highlights a new trend: cybercriminals leveraging AI to make ransomware smarter and more dangerous. The big question now is—are our current security measures ready for this next-gen threat?
Perplexity has launched a new, transparent revenue share program for publishers, aiming to compensate media outlets when their content is used or referenced in AI-powered search results and tasks.
How Perplexity’s Revenue Share Works
Perplexity’s initiative introduces a subscription tier called Comet Plus, priced at $5 per month, that enables publishers to earn money through three channels:
Users will now be directed straight to publisher websites via the Comet browser.
When publisher content is cited in search answers.
When the AI agent uses content to fulfill user tasks. Unlike traditional ad-supported models, this program pools subscription revenues and distributes 80% directly to publishers, with Perplexity keeping 20% for operational costs.
Funding and Payouts
The program starts with an initial $42.5 million revenue pool dedicated to publisher compensation, with plans to grow this fund as subscriptions increase. Payments will be calculated based on how frequently a publisher’s content is used in AI search results, cited in answers, or utilized by AI agents.
Industry Impact and Reception
This model is one of the first by an AI startup to transparently share revenues based on actual content usage, representing a shift from the multimillion-dollar exclusive licensing deals common with AI giants like OpenAI and Google. This move comes as a response to criticism that AI tools exploit journalism without fair return and is intended to set a more balanced compensation model in the age of AI.
Publisher Participation
Perplexity is actively seeking major publishing partners and has previously worked with outlets like TIME, Fortune, the LA Times, Blavity, and Der Spiegel. While specific new partners for Comet Plus haven’t been disclosed, the company emphasizes that the program is open to collaboration and feedback from the media industry.
The Fresh Take
With the launch of Comet Plus and this revenue-share model, Perplexity is attempting to set a precedent for sustainable media monetization in the AI-driven web. It emphasizes transparency, direct payouts based on actual usage, and an open invitation to publishers to help refine the system—moving beyond ad clicks and traffic, toward a business model designed for the modern search experience.
Meta just dropped an exciting update for AI enthusiasts—it’s partnering with Midjourney to integrate their cutting-edge aesthetic technology into Meta’s future AI products. The goal? To make AI not just smart, but visually beautiful.
According to Meta, this collaboration is more than just a licensing deal. It’s a technical partnership where both research teams will work closely to push the boundaries of AI creativity. And honestly, it makes sense. Midjourney has already set the bar high in the world of AI-generated art, so combining that with Meta’s scale could be game-changing.
Meta is going all-in to deliver top-notch products—bringing together world-class talent, huge computing resources, and strategic partnerships with leading industry players. With Midjourney’s visual flair and Meta’s ambitious roadmap, this could reshape how we experience AI-generated content across social media and beyond.
The companies are keeping details under wraps for now, but they promise more updates soon. If this is just the beginning, the future of AI visuals might be a whole lot more beautiful.
Elon Musk is back in the spotlight, and this time it’s not about rockets or EVs. His AI company, xAI, just made a big announcement – Grok 2.5 is now open source. Yep, the same advanced language model that’s been grabbing attention is now free for developers and researchers to play with.
The @xAI Grok 2.5 model, which was our best model last year, is now open source.
Well, Grok 2.5 isn’t just another chatbot brain. It’s built to compete with the big guns like GPT and Google’s Gemini. By opening it up, xAI is giving everyone – from indie coders to tech giants – the chance to experiment, tweak, and build on top of a powerful model without paying hefty licensing fees.
What Makes This Different?
Here’s the interesting part: most AI companies keep their models under lock and key. But Musk is taking a different route, betting on transparency and community-driven progress. This move not only levels the playing field but also speeds up innovation in areas like chatbots, enterprise tools, and AI-driven apps.
Why Would Musk Do This?
Probably to challenge the growing dominance of closed AI ecosystems. Plus, let’s be honest, open-source projects often grow faster because thousands of brains work on them instead of just one company’s team.
What Does It Mean for Developers and Businesses?
For developers, this is honestly a big deal. Imagine building apps with serious AI power without burning through your budget – that’s what this unlocks. Businesses can finally tinker with custom AI tools without being stuck paying crazy fees to big vendors.
The Bigger Picture
And for the AI world? It feels like a push toward making this stuff way more accessible to everyone, not just big tech. It’s not just xAI making bold moves. OpenAI has also opened parts of ChatGPT to the public. Two new open-weight models—gpt-oss-20B and gpt-oss-120B—are now available under the Apache 2.0 license. The smaller one runs on laptops, while the bigger model works on desktops or cloud GPUs. This marks a clear trend: even AI giants are embracing openness, and xAI’s Grok 2.5 release shows the competition is heating up.
The Takeaway
Grok 2.5 going open source isn’t just another update in the news feed – it’s a bold move that could seriously shake up the whole AI race.
When OpenAI announced GPT-5, the internet went wild. Threads on Reddit, endless tweets, and tech influencers all hinted at one thing: this was going to be the upgrade. People imagined an AI so advanced that it could practically run businesses, write flawless code, and maybe even predict your next thought.
The reality? GPT-5 is good, no doubt, but it’s not the earth-shattering leap that marketing made it sound like. Here’s why the buzz was so big — and why expectations didn’t quite match reality.
Why GPT-5 Created So Much Noise
OpenAI knows how to build hype. Before launch, the company teased big phrases like “PhD-level intelligence” and introduced a new feature called Agent Mode. Social media picked it up, and soon the narrative was clear: GPT-5 was going to change everything.
People started imagining an AI that would not just answer questions but act like a personal digital employee — booking meetings, managing tasks, even running your projects. Add to that the usual chatter about “the next step towards the future,” and expectations went through the roof.
The Real Improvements
Now, let’s be fair. GPT-5 is definitely better than GPT-4. It brings:
Stronger reasoning skills: Great for complex, multi-step tasks.
Better context retention: Conversations feel smoother without losing the thread.
Model routing: GPT-5 can choose between models for speed and accuracy.
Agent Mode (beta): Designed to break big tasks into steps and execute them.
For power users — developers, analysts, and advanced researchers — these upgrades matter a lot. If you need AI for logic-heavy work, GPT-5 is a solid step forward.
So, Where Did It Fall Short?
For everyday users, the change isn’t dramatic. Writing emails, summarizing docs, or brainstorming ideas? GPT-4 already handled those well, and GPT-5 doesn’t feel like a game-changer here.
The most hyped feature, Agent Mode, is still rough around the edges. It’s slow, sometimes overcomplicates tasks, and feels more experimental than production-ready. Great idea, but not yet the AI assistant that replaces half your apps.
Another challenge? Expectation management. Marketing set the bar so high that anything short of magic was going to feel disappointing.
The Verdict: A Step Forward, Not a Leap
GPT-5 is not a flop. It’s a solid, thoughtful update that improves reasoning, context, and efficiency. But it’s not the “mind-blowing” jump some hoped for. Think of it as part of a marathon, not a sprint.
If you’re into deep problem-solving, GPT-5 is worth trying. If you were expecting a miracle worker, you might want to temper your expectations. The AI race is heating up, and GPT-5 is another sign that progress is real — just not overnight.
https://www.youtube.com/watch?v=PlP7QNPEkQARead more on: OpenAI, Fortune
Meta just rolled out an AI-powered voice translation feature for Instagram and Facebook Reels. It lets creators easily dub their videos into English or Spanish, matching their lip movements for a natural look. This means creators can reach new fans worldwide without language barriers. The tool is live for public accounts with 1,000+ followers, but not everywhere yet—some countries and U.S. states are missing out for now. Creators can preview translations and check how each language performs. Just face the camera and speak clearly for the best results. This could change how content travels across the globe.
Eight Sleep just scored $100 million to power up its smart sleep gear and step onto the world stage. This startup uses AI to create digital sleep twins that help predict and improve rest. They’re also moving into medical tech, aiming to tackle sleep apnea and menopause symptoms with smart cooling. Since 2019, they’ve sold over half a billion dollars in sleep pods and are now shipping to 30-plus countries. With this fresh cash, Eight Sleep wants to get even smarter and expand into China soon. Sleep better, dream bigger!
Grammarly introduced eight AI-driven writing assistants and a redesigned document interface on August 18, transforming from a grammar tool into a productivity platform. These agents help students and professionals with grading, citations, reader insights, and expert reviews while maintaining authentic authorship. The update addresses the growing AI literacy gap, as many students feel unprepared for AI in the workforce. Built on technology from Grammarly’s recent acquisition of Coda, the new block-based design includes an AI Chat assistant. The rollout begins with Free and Pro users, with Enterprise and Education access coming soon.
Amid rising unemployment among young tech professionals, industry leaders are divided over the role of AI in entry-level hiring. AWS CEO Matt Garman criticized replacing junior staff with AI, calling it shortsighted and warning about future talent gaps. Contrarily, OpenAI and Google leaders foresee AI soon replicating junior roles. Reports show new graduate hiring in tech has dropped sharply, with recent grads now a small fraction of new hires. Garman urges students to build adaptable, creative skills, while Amazon signals more workforce cuts, fueling the debate over AI’s place in the workforce.
