22 Weekly AI Updates: OpenAI's $1.1B Deal + ChatGPT New Features + Krea's Real-Time Video AI

This week, 160 plus AI updates flooded the market. But these 22 top stories will rewrite the entire industry. Let's get started. Sam Altman and OpenAI just

made a huge move by acquiring this Indian company Statig founded by Vijay Raji, an entrepreneur who just made $1. 1 billion. With this acquisition, Raji is now the CTO of applications at OpenAI, meaning he's leading the future development of chat GPT for billions of users worldwide. Raji isn't just another founder. He built the technology behind AB testing for giants like Meta, Airbnb, and Notion. So, every time Instagram or Facebook tries a new feature, it's Raj's Tech making that happen. Now, here's where it gets really exciting. OpenAI didn't just acquire a company. They bought the ability to move 10 times faster. As CTO of applications, Raji will oversee the development of Chat GPT and Codeex, testing thousands of features at once. This means we'll be seeing weekly improvements instead of waiting months for updates. While other tech giants like Google and Anthropic are spending billions to train bigger models, OpenAI's focus is on speed. Raji's methodology ensures rapid iteration getting real user feedback before fully rolling out new features. This is a massive win for Indian innovation and shows how talent from India is reshaping the global AI landscape. Raji's expertise is now at the core of OpenAI strategy proving how India is leading in the tech world. Open

AAI just made four major moves that totally reshape their platform strategy. First up democratizing AI. OpenAI made chat GPT projects available to all free users. These are AI powered workspaces that keep context across conversations related chats, uploaded files and custom instructions all in one place. Free users get five file uploads per project plus users get 25 and enterprise users get 40. They also rolled out memory

improvements to free users. What used to be behind payw walls is now available to everyone. Next, the workflow revolution. On September 4th, they launched conversation branching. Now you can hover over any message, click branch in new chat and explore different approaches without losing your main thread. Imagine marketing teams branching one brainstorming session into formal, casual and humorous versions, comparing them side by side. It's like version control for ideas finally solving the issue of overwriting or starting from scratch. Third, a power tool for prompt optimization. On August 7th, they launched GPT5's prompt optimizer. Now you can paste any prompt, click optimize and get systematic improvements with explanations. Finally, the strategic strike. Open AAI announced the OpenAI jobs platform, a LinkedIn competitor launching in mid 2026, powered by AI job matching. The twist

LinkedIn is owned by Microsoft. OpenAI's biggest backer with $13 billion invested. OpenAI is partnering with Walmart to certify 10 million Americans in AI skills by 2030. This puts them in direct competition with Microsoft's professional networking monopoly. Sam Alman revealed this during a White House tech dinner, making it clear OpenAI is going after Microsoft stuff. The pattern here is clear. OpenAI isn't just building better AI, they're creating the infrastructure to take on big tech across multiple fronts. Forget everything you know about video creation. Create AAI just dropped

something that feels like magic. You can paint on a canvas, type a prompt, or even use your webcam and watch AI generate video in real time at 12 plus frames per second. All while keeping everything perfectly consistent. This isn't your average AI video tool. Behind the scenes, Korea uses world model technology, which means the AI actually understands how scenes evolve and how actions ripple through time. The result, true responsiveness with no loss in visual coherence. Each video is capped at 10 seconds, but it renders faster than playback speed, so you get instant feedback as you edit. Want to try it out? Head over to krea. ai/realtime and have some fun with it. People on Twitter are already freaking out over this. And honestly, it's easy to see why. Google Photos just got a serious upgrade with V3, and now your pictures

come to life. This new feature available in the create tab of the mobile app for US users takes any still image and turns it into a short video clip. Google Photos had basic animation before with V2, but VO3 brings a whole new level of quality. The old version had subtle movements or a surprise animation with the I'm feeling lucky feature. But with V3, you're getting smoother, more natural looking motion in your photos. The best part, this feature is still free, but there's a limit to how many times you can use it. For those with AI Pro and AI Ultra subscriptions, you get access to more generations. However, the V3 version doesn't include audio, and videos are capped at 4 seconds. This is Google bringing its latest AI video tech straight to consumers through Google Photos, a product with 1. 5 billion monthly active users. That's a huge advantage when it comes to showing off V3's capabilities on a massive scale. Elon Musk's XAI just dropped Gro Code Fast One and the numbers are absolutely

crushing the competition. This isn't just another coding model. It's built specifically for agentic coding workflows. Fast one runs at a blistering 160 tokens per second. For context, GPT5 hits 50. 1 tokens per second. Gemini 2. 5 Pro does 92. 4 and Claude for Sonnet comes in at 78. 7. That's more than three times faster than OpenAI's flagship model. On the SWE bench verified benchmark, Grocode fast scored 70. 8%. Competitive with GPT5 74. 9% and Claude Sonnet Force 72. 7%. But it does it at lightning speed. They're offering it free for a limited time through platforms like GitHub Copilot, Cursor, Klein, and more. One developer said, "Grock code fast is ridiculously fast. I had to change up how I work in cursor. I threw together a battle simulator prototype in less than a day. I found I'm better off giving it smaller, focused tasks and iterating quickly. " GitHub's Mario Rodriguez put it this way. In early testing, Grock Codefast has shown both its speed and quality in energetic coding tasks. This is a compelling new option for our developers. Anthropic has raised $13 billion, bringing its valuation to $183

billion, nearly triple its worth in March, putting it in the same league as Microsoft, Meta, and Google. This leap positions Anthropic as a major player in the tech world, transitioning from a safety first AI startup to a corporate giant. Top tier investors like Iconic, Fidelity, and Lightseed are backing Anthropic. These firms are not just betting on potential. They're betting on Anthropic overtaking OpenAI. With $5 billion in projected revenue by August 2025, Anthropic is one of the fastest growing tech companies, surpassing companies like Uber and Airbnb in valuation. Warner Brothers just filed a massive lawsuit against Mjourney on September 4th, and the details are

explosive. They accused Midjourney of generating images and videos featuring iconic copyrighted characters like Superman, Batman, Bugs Bunny, and Scooby-Doo without permission. This makes Warner Brothers the third major studio to go after Midjourney, joining Disney and Universal, who filed their suits in June. But Warner Brothers complaint is particularly aggressive. They're asking for up to $150,000 per infringed work for willful infringement. What's even more shocking is this. Warner Brothers claims Midjourney recently removed guardrails that previously prevented the AI from generating infringing content. When Midjourney first launched video generation, they blocked many infringing images, but within the last few weeks, they allegedly removed those protections and even called it an improvement. The lawsuit includes sidebyside comparisons of Midjourneys outputs next to actual Warner Brothers characters. Even simple prompts like classic comic book superhero battle are generating downloadable images of Superman, Batman, and Flash. Warner Brothers is also calling out Midjourney TV, a 247 streaming channel launched in July. They claim it's directly competing with their content using their own characters. And the numbers, Midjourney has nearly 21 million subscribers, pulled in $300 million in revenue in 2024. Atlassian just made a massive move. They're acquiring the browser company for a cool

$610 million in cash. This is the team behind the Arc and Dia browsers, and they're all in on building the AI browser for knowledge work. Mike Cannon Brooks, Atlassian CEO, put it perfectly. Today's browsers weren't built for work. They were built for browsing. This is a bold step forward in reimagining the browser for knowledge work in the AI era. Here's the deal. The browser company will keep operating on its own, but their main focus is going to be DIA, the AI powered browser. The idea is to create a browser that's optimized for all the SAS apps that knowledge workers use everyday and pack it with AI features and work memory. Sounds like the future, right? This is a huge play for Atlassian, which already serves over 300,000 customers, including 80% of the Fortune 500. They're now in a perfect position to introduce DIA to millions of enterprise users who already use Jira, Confluence, and their other tools. Josh Miller, the CEO of the browser company, said this deal is going to help them move faster and dream bigger. Building an AI browser that's trusted by companies, but still feels personal for each user. This is a huge one. Anthropic just agreed to pay $1. 5 billion to

settle a class action copyright lawsuit, the largest settlement in US history. This news dropped on September 5th, right after Anthropic raised $13 billion at an $183 billion valuation. Here's the story. Authors sued Anthropic for downloading millions of pirated books from shadow libraries like Library Genesis and Pirate Library Mirror to Train Claude. We're talking about 500,000 published works. Now, the settlement is $3,000 per book plus interest. But the real twist, in June, federal judge William Alop ruled that while training AI on copyrighted material is fair use, pirating it isn't. That's theft. So, this settlement isn't about AI training being illegal. It's about the piracy. Anthropic has to destroy all the pirated data sets they took. Justin Nelson, the attorney handling the case, called it the first of its kind in the AI era. He says it sends a clear message to companies don't steal from pirate sites. And this is big for the whole industry. Meta is facing the same issue with LibGen and OpenAI has its own copyright cases to deal with. For Anthropic, this settlement is doable. But for smaller startups, this could be devastating. Imagine this. You're lying in a hospital bed, fully aware, but unable to move or speak. Your

family thinks you're gone, but you're desperately trying to respond to their voices. It's a nightmare scenario, but this breakthrough might just change everything. Researchers at Stony Brook University developed CEI, an AI that detects consciousness in coma patients by analyzing facial movements down to the level of individual pores. The crazy part, when doctors gave simple commands like, "Open your eyes," the AI detected responses an average of 4. 1 days before human clinicians noticed anything. In a study of 37 comeomaos patients, CMI detected eye movements in 30 out of 36 cases and mouth movements in 16 out of 17 videos. But here's where it gets even more interesting. Columbia University found that patients with specific brain wave patterns called sleep spindles had a 76% recovery rate. So, you've seen the major stories this week. Now, let's rapidly blast through the fastest and wildest new AI advancements that just launched. Ready? Let's go. Apple's fast VLM just went live for browser testing, delivering 85 times faster video captioning than competing models. This accessibility breakthrough runs entirely on device with zero cloud dependency and works on everything from iPhones to Raspberry Pi. 11 Labs dropped sound effects v2 with seamless looping capabilities and extended duration up to 30 seconds at 48 kHz quality. Studiograde audio from simple text prompts. No expensive licenses needed. Hen's Avatar 4 turns any single photo into a talking avatar with multilingual support across 175 plus languages. The AI interprets vocal tone and emotion to generate photorealistic facial movements in real time. Amazon launched Lens Live on September 2nd, letting you point your camera at anything and instantly buy matching products. Nuts. It's live object detection with one tap purchasing, currently rolling out to iOS users. Google Circle to search now supports continuous translation as you scroll in over 70 languages. No more restarting the translation process every time content changes on your screen. Mistrial threw down with a massive move. They gave away 20 plus enterprise integrations and memory capabilities for free, directly targeting OpenAI's paid model and undercutting chat GPT plus pricing strategy. Notebook LM expanded beyond podcast generation, now creating interviews, debates, and educational content from any document upload, making it even more powerful for researchers and creators with four distinct audio formats. WordPress launched TX at WordCam US 2025, automating Gutenberg block creation for WordPress sites and saving developers hours of manual coding work. Firecrawl introduced an AI readiness checker that scans websites and provides optimization checklist. Think SEO audit, but for the AI era. Whimo expanded testing to Seattle and Denver this week, though they're starting with human drivers behind the wheel before moving to fully autonomous operations in these new markets. Humanoid robots hit a major milestone with figure 02 loading entire dishwashers with fingertip precision using the same Helix AI model that previously mastered laundry folding and package sorting. Tencent open- sourced Huan Video Foley generating Hollywood level sound effects perfectly synced to video at professional 48 kHz quality. The silent AI video era just ended with this game-changing release. Which update caught your attention? Let me know in the comments. Join our WhatsApp channel and don't forget to subscribe for next week's fire update.