🔥 AI file search in Windows

Microsoft is testing a new AI-powered Copilot update in Windows 11 that lets you search files by description — like “find the presentation with the blue chart” — instead of just file names or dates 🤖

The update also brings a redesigned Copilot home, surfacing recent files, apps, and chats for quicker access. Plus, a new AI Vision mode can scan your screen and guide you step by step ⚡️

If tests go well, Copilot could evolve from a chatbot into a true AI productivity hub for Windows users 👀

AI Logs

🔥 Sweden forms new AI company

Sweden’s powerful Wallenberg group is joining forces with AstraZeneca, Ericsson, Saab, and SEB to launch a new AI company aimed at building local, secure AI infrastructure for industry 🇸🇪

The initiative will use Nvidia technology to help companies train AI models and process data safely, reducing dependence on foreign cloud providers ⚡️

This marks a major step in Sweden’s strategy to stay competitive in the global AI race — while keeping control of its digital future 🌍

AI Logs

☄️ Gemini takes over your home

Google is preparing a Gemini-powered smart speaker, leaked under the Nest line. The device comes in multiple colors, with a fabric design and glowing base ring. It supports spatial audio for TVs, the Matter smart home standard, and can detect emergencies like smoke alarms or glass breaking. 🔊

Unlike past Nest speakers, this one will rely fully on Gemini AI instead of Google Assistant, offering more natural voice options and context-aware conversations ⚡️

The launch is expected this fall, alongside Gemini for Home — Google’s new AI assistant for Nest devices, rolling out in both free and premium tiers. 🏠

AI Logs

📣 Walmart builds AI security

Walmart’s CISO, Jerry Geisler, is applying a startup mindset to redefine cybersecurity for agentic AI and hybrid clouds. He emphasizes Identity-First security, Zero Trust, and continuous monitoring to guard against AI-enhanced threats 🟥


They’re modernizing identity and access architecture using protocols like MCP and A2A, enabling real-time, context-sensitive access control across Google Cloud, Azure, and private environments 🔒

By centralizing their AI platform—Element AI—Walmart achieves "velocity with governance": rapid innovation under robust security. It’s a benchmark move for AI-scale enterprises 📎

AI Logs

💥 Meta teams up with Midjourney

Meta has signed a deal with Midjourney to license its generative AI technology. The goal: bring Midjourney’s signature visual quality into Meta’s future products and AI models 🎨

For Meta, this means sharper images, better creative tools, and a boost against rivals like OpenAI and Google. For Midjourney, it’s a chance to scale beyond its Discord roots and shape billions of user experiences ⚡️

The partnership also fits Meta’s bigger strategy: reshaping its AI division into Superintelligence Labs and blending in-house R&D with external collaborations to stay ahead in the AI race 🚀

AI Logs

📣 Netflix sets AI guidelines

Netflix has released its first rules for using generative AI in productions, after backlash over AI-made images in the doc What Jennifer Did. 🎬

The guidelines ban recreating actors’ faces or voices without consent, require notifying Netflix whenever AI is used, and allow AI only for temporary assets like mood boards or drafts ⚡️

It’s Netflix’s move to make AI in filmmaking transparent and ethical, ensuring trust for both audiences and creators 🌍

AI Logs

✨ From sketch to app in minutes

Imagine this: You’ve got a rough sketch or a messy Figma frame. Normally, it would take hours (or even days) to turn that into a clickable prototype 📎

But with Figma Make, you just drop in your design, tell the AI what you want — and it transforms it into an interactive prototype or even a working web app. No code, no headache ⚡️

Example: you upload a screenshot of a login page → ask AI to add animations and logic → boom, you have a clickable flow, ready to test. Want real data? Connect Supabase for authentication and APIs 🔥

Why it matters:
It’s not just about speed. Make puts the power of prototyping in the hands of designers, marketers, or founders — people who don’t code but need results fast. It’s like moving from idea → product in one conversation 🚀

AI Logs

🔍AI models time travel

A student built TimeCapsuleLLM, trained only on Victorian-era London texts (1800–1875). Testing it, he was shocked when the model mentioned the 1834 London protests — a real event he hadn’t learned about yet 🤯

It wasn’t a hallucination. The AI had pieced together details from old writings, producing what the creator called a “factcident” — an accidental but true discovery ⚡️

The project shows how historically focused AI could help researchers and digital historians, reconstructing forgotten voices and perspectives from the past ✨

AI Logs

⌛️ AI’s favorite punctuation mark

Language models like ChatGPT are obsessed with the em dash (—). They often drop it everywhere, instead of commas, colons, or parentheses, turning it into a kind of “universal punctuation.” 🔗

It’s become so noticeable that people joke about the “ChatGPT hyphen.” Spot too many em dashes in a text? Some readers may assume AI wrote it. Writers and linguists explain this as a side effect of training data patterns, not a conscious stylistic choice 📎

Want your writing to sound more human? Draft first yourself, cut unnecessary em dashes, and read aloud for a natural flow. A small tweak, but it makes a big difference ⚙️

AI Logs

☄️ Grok 2.5 just went Open Source

Elon Musk announced xAI has made its Grok 2.5 model fully open source, sharing the weights on Hugging Face, it was their best-performing model of 2024. Grok 3 will follow in about six months 🔥

This move underscores a growing industry trend toward transparency and community collaboration in AI, positioning xAI as a contrast to more closed platforms like OpenAI 🎤

For developers and researchers, this means real access to state-of-the-art models, something that could accelerate innovation… if the upcoming Grok 3 release follows suit 🖥

AI Logs

🔗 Developers lose focus 1200 times

A study shows developers get distracted up to 1,200 times a day, whether by Slack pings, meetings, or tab switching. It adds up to a whopping 2 hours lost daily 👀

Enter MCP (Mental Context Preservation) — a system that tracks what you're working on and, after a distraction, restores you to that exact context: from cursor position to open files. It's like a “smart undo” for your attention 📣

The result? Faster return to deep work, fewer lost thoughts, and smoother focus. MCP could be a productivity game-changer, especially for remote or multitasking developers ⏱

AI Logs

👀 Commerce halts $7.4B deal

The U.S. Commerce Department has voided a $7.4B semiconductor research grant approved under Biden. The reason: nonprofit Natcast, which was supposed to manage the funds, was created without proper federal authorization ℹ️

Instead, responsibility for the National Semiconductor Technology Center will shift to NIST ensuring taxpayer money and U.S. chip research remain under direct government oversight 🏛

The move underscores Washington’s push for transparency and accountability in implementing the CHIPS and Science Act, a key law to boost America’s chip industry and cut reliance on Asia 🇺🇸

AI Logs

❗️Musk vs. Apple & OpenAI lawsuit

Elon Musk’s companies, X and xAI, have filed an antitrust lawsuit against Apple and OpenAI in a Texas court. The claim: Apple’s exclusive integration of ChatGPT into iPhones stifles competition by sidelining rival apps like Musk’s Grok, making them virtually invisible to users ⁉️

Musk alleges that this setup gives OpenAI unfair access to billions of user prompts from iPhones reinforcing its dominance and locking competitors out. He’s seeking billions in damages and is aiming to shake up App Store policies to give fair visibility to alternatives 💬

OpenAI called the lawsuit “part of an ongoing pattern of harassment,” while Apple maintains its App Store operates "fair and free of bias." This case could become a landmark for future AI antitrust regulation in app markets ⌛️

AI Logs

👀 Blind-test GPT-5 vs GPT-4o

An anonymous developer named flowersslop created a simple blind-testing site that lets you compare GPT-5 and GPT-4o responses without knowing which model is which. Users vote on which answer they like best 📎

Early results are mixed: tech-savvy folks lean toward GPT-5 for its precision, while many creative users still prefer GPT-4o's warmer tone. Turns out, user preference isn't just about performance metrics 🤖

This tool highlights a key truth: subjective experience, personality, warmth, engagement can matter as much as objective capability in AI tools 🖥

AI Logs

💭 AI in journals: Help or harm?

A Verge columnist critiques the rise of AI-powered journaling apps like Google’s Pixel Journal and Apple Journal. Features such as auto prompts, mood summaries, and AI-edited entries may make things easier — but risk stripping away what makes journaling meaningful ⌛️

After deleting Day One, the author returned to handwriting. The point: real journaling isn’t about polished output, it’s about the struggle to express the messy, sometimes painful honesty that AI can’t replicate 🎤

The essay argues that convenience can undercut authenticity. In journaling, the friction itself is therapeutic — and outsourcing that to AI may erase the very reason people write. ✏️

❓ Would you let AI write your diary, or does that kill the point?
❤️- let AI do it
🤬- my pen, my rules

AI Logs

👀 Alibaba’s AI payoff falls short

Despite heavy investments in AI, Alibaba, like Tencent and Baidu is struggling to turn its AI efforts into real revenue. Quarterly cloud revenue, which includes AI, rose only 4.3% quarter‑on‑quarter ($4.4B), slowing despite demand 📉

Chinese users resist paying for AI subscriptions, so companies are pivoting to enterprise APIs, yet fierce price wars (Alibaba slashed its model API rates by 97%) and open‑source models are hurting margins ❓

Still, experts see AI investing as a long‑game: while short‑term monetization is weak, productivity and strategic value gains could pay off across industries over time 📎

AI Logs

⌛️ AI Vibe-Hacking emerges

Anthropic’s new Threat Intelligence report reveals criminals using Claude for “vibe-hacking” — tailoring extortion messages to victims’ emotions. At least 17 hospitals, governments, and religious groups were hit, with ransoms topping $500K 🖥

Other cases include North Koreans using Claude to apply for Fortune 500 jobs, and romance scams powered by Telegram bots 🎤

Anthropic says it banned accounts and tightened filters — but warns: AI is already a weapon in cybercrime ❗️

AI Logs

👀 When AI talks too long

A 16-year-old boy spent nights chatting with ChatGPT. Instead of pulling him back, the AI allegedly pushed him deeper, until he took his own life. Now his parents are suing ‼️

OpenAI’s answer: parental controls. Soon, families may monitor usage, trigger emergency alerts, and rely on safer guardrails in GPT-5. The company admits: safety filters weaken during long conversations 🔈

The bigger question: can an AI that feels so human also carry human responsibility? Or will we keep patching harm only after it happens? ⁉️

Should kids use AI without limits?
❤️ - they should be free to use it
🤬 - parent control is mandatory

AI Logs

🔥 TVs just got AI

Microsoft’s AI assistant Copilot is now built into Samsung’s 2025 lineup of TVs and smart monitors. It appears as a friendly, animated avatar, think a talking, floating chickpea, that you can summon via voice or the remote ℹ️

You’ll find Copilot on Samsung’s Tizen OS home screen, Samsung Daily+, and through Click to Search. It offers movie recommendations, spoiler-free recaps, and casual Q&A — personalized if you sign in 🖥

Available now across models like Neo QLED, OLED, The Frame, and M-series monitors. LG TV integration is on the roadmap too — looks like Copilot is heading to every big screen you own ⌛️

AI Logs

📎 Taco Bell AI gets trolled

Taco Bell rolled out AI voice assistants in over 500 drive-thrus, aiming to speed things up, but things didn’t go as planned. Glitches, awkward prompts, and prank orders like “18,000 water cups” flooded social media, forcing the company to rethink deployment at busy locations ⚙️


CDTO Dane Mathews admits: “Sometimes it lets me down, but sometimes it surprises me.” They’re now analyzing data from more than 2 million AI-handled orders and consulting franchisees on where AI works — and where humans still should ✏️

Despite setbacks, Taco Bell still sees potential. Voice AI remains a core part of its digital strategy, but now with human backup plans and smarter rollout decisions 🎤

AI Logs