Good morning!

Today’s AI headlines tell an interesting story.

First, we have new models, new capabilities, and an increasingly crowded race for leadership. China is developing its own AI chip, a new trillion-parameter model is on the way, and tomorrow we’ll (hopefully) get the long-awaited GPT-5.6.

The competition is no longer just about building the best model. It’s increasingly about pricing, positioning, and ecosystems. Have you heard of Token Unmaxxing yet? If not, you probably should.

Meanwhile, AI continues to spill into everyday life, raising important questions about digital independence. Take Ukraine, for example: it’s investing in self-hosted AI models to reduce its dependence on the shifting priorities and policies of Uncle Sam.

Let’s unwrap them!

🚀 New models, new capabilities, and an increasingly crowded race for leadership.

• Behind the Delivery Riders: Meituan Quietly Built a Trillion-Parameter AI Model (pandaily.com)

• Meta’s latest AI model can generate images, websites and even QR codes (livemint.com)

• China’s DeepSeek developing its own AI chip, sources say, China News (asiaone.com)

• OpenAI to unveil GPT-5.6 on Thursday after delaying launch (brecorder.com)

• SpaceXAI: SpaceXAI plans to launch new model with Cursor as soon as Wednesday (economictimes.indiatimes.com)

• Anthropic Extends Free Access to Claude Fable 5 on All Paid Plans (cybersecuritynews.com)

⚔️ The battle is now about pricing, positioning, ecosystems and market strategy.

• Token Unmaxxing: AI’s Price War Is Starting to Reshape the Trade (uk.investing.com)

• OpenAI and Anthropic are pulling in different directions (helpnetsecurity.com)

• Token Maxxing: This Pricing Movie Is A Remake (abovethelaw.com)

🌍 From entertainment to search and even geopolitics, AI keeps spilling into everyday life.

• ByteDance, Tencent, and Alibaba Are Shutting Down AI Chat Companions (pandaily.com)

• “This is humans versus machines, and today, the humans are rising”: The tribute group formed to rival AI band Velvet Sundown (musicradar.com)

• Generative Engine Optimization isn’t the new SEO. Here’s what brands need to do instead (diginomica.com)

• Ukraine Pushes for Self-Hosted AI Models (thegaze.media)

🛡️ As AI grows more powerful, the hardest questions are becoming human rather than technical.

• Hackers can use 9 of the most popular AI tools to assemble massive botnets (arstechnica.com)

• Can AI models consent to their own constitutions? (marginalrevolution.com)

• MEGA Army App Recognizes All Combat Vehicles with Full AI-Based Technical Analysis (armyrecognition.com)

• Anthropic Removes Hidden Claude Code Tracker After Researchers Raise Privacy Concerns (decrypt.co)

Share

🧠 AI isn’t just changing technology—it is quietly reshaping how we think, create and relate to each other.

• The Mind I Said AI Didn’t Have (psychologytoday.com)

• AI-generated ‘actress’ Tilly Norwood making feature film debut in Misaligned (abc13.com)

• People Who Use AI More Also Dislike It Most, Study Reveals (novaramedia.com)

• Experts advise journalists on ethical use of AI (gazettengr.com)

• Philosophy of Mind Cannot Explain Artificial Intelligence (psychologytoday.com)

🛡️ As AI becomes infrastructure, the questions shift from what it can do to who can trust it.

• Automated Moderation Is Here to Stay (eff.org)

• AI Sovereignty Is a New Test for Enterprises (databreachtoday.co.uk)

• China May Restrict Access to Its Most Powerful AI Models (time.com)

• Resilience by Design: Preparing Your AI Stack for an Era of Uncertainty (bain.com)

• The Safeguard That’s Eating Itself (techuk.org)

💼 The workplace is becoming AI-native—whether employees are ready or not.

• Why AI-built tools are threatening SaaS vendor renewals (informationweek.com)

• Claude Cowork expands to mobile and web (techcrunch.com)

• Claude Cowork can now keep working even after you close your laptop ( (digitaltrends.com)

• Workplaces Have Gotten So Bizarre That People Are Just Sending AI Slop Back and Forth at Each Other (futurism.com)

• The AI productivity trap in enterprise finance (techfinitive.com)

• How To Use ChatGPT To Remember Almost Anything (forbes.com)

• Everyone Is Using AI To Accomplish Their Goals. Here’s How You Can, Too (forbes.com)

• Job seekers in Czechia are hiding secret AI prompts in their CVs: Does it work? (expats.cz)

🖥️ Running AI on your own hardware is becoming less of a hobby and more of a viable alternative.

• AI Fast-Tracks Open Web’s Collapse (mediapost.com)

• My local LLM can call every tool that Claude can, except it runs on my own hardware (xda-developers.com)

• I run a 32GB GPU instead of paying for Claude or Codex, and Qwen 3.6 keeps up more than I expected (xda-developers.com)

Share

🧠 AI keeps drifting into strange cultural and cognitive territory—less tool, more environment.

• A next-generation religious chatbot explores how faith and AI interpretation intersect (timesofisrael.com)

• “AI poisoning” emerges as a term for corrupting or manipulating model behavior and training data (digiday.com)

• Yann LeCun advances research into more flexible AI architectures beyond current dominant approaches (bbc.com)

⚠️ The attack surface is expanding faster than the defenses.

• Malware targeting Claude Code and OpenAI Codex uses scanner-evasion techniques to avoid detection (cybersecuritynews.com)

• A 15-year-old is arrested following a reported ChatGPT-assisted cyberattack on Bandai Channel systems (cybersecuritynews.com)

• UK officials warn that AI now represents the most significant security challenge of the decade (insurancejournal.com)

🌍 The real world is already adapting—sometimes badly, sometimes unevenly.

• Chinese tech firms restrict AI-generated web novels after backlash from readers (restofworld.org)

• Evidence grows that AI-assisted learning may be linked to declining math performance in students (hechingerreport.org)

• Companies and educators debate who should be responsible for teaching AI skills in the workplace (businessinsider.com)

• AI-driven “casino logic” is reshaping how competition and strategy are framed in tech industries (cio.com)

• A new report argues we are approaching a tipping point in local AI deployment and adoption (neoteric.no)

🧭 The real conflict is shifting from capability to governance, ownership, and resistance.

• Sam Altman calls for a global AI regulator while also suggesting U.S. government financial involvement in OpenAI (forbes.com)

• A growing movement argues that AI should be actively resisted or stopped, not simply regulated (thewalrus.ca)

Share

AI stories worth your attention today.

🛠 Tools

• Google Health CLI brings AI agents to your Fitbit and health data (androidauthority.com)

• Microsoft struggles to address AI notetaker governance nightmare (computerworld.com)

💡 Use Cases

• I asked ChatGPT to plan my weekends for a month (siliconcanals.com)

• These small Claude Code hooks turned my assistant into something I actually wanted to keep (xda-developers.com)

🧠 Psychology

• Study finds humans will talk to AI ghosts of the dead as reincarnations (digitaltrends.com)

💼 Business

• OpenAI to Offer Trump Administration a 5% Equity Stake and Other AI Firms Could Also Follow (benzinga.com)

• Why businesses are choosing cheap Chinese AI models over AI giants (business-standard.com)

• Indian tech tycoon bets $30M of his own money to build AI alternative to Microsoft Office (techcrunch.com)

⚖️ Policy

• Europe looks to fight any forced shutdown of AI (computerworld.com)

• You Own What Your Chatbot Says (pymnts.com)

• Godot bans “autonomous AI agent use or vibe coded” contributions (gamesindustry.biz)

🔬 Research

• Anthropic’s Claude Is Growing 9 Times Faster Than ChatGPT (benzinga.com)

• OpenAI and Anthropic Built AI Around Tokens. Palantir’s Karp Says ‘Something Has Gone Completely Wrong’ With The Model. (ibtimes.com)

⚠️ Risks

• “Digital nuclear weapons” - CIA puts artificial intelligence in the zone of extreme risks (bursa.ro)

• Top tech companies are researching whether AI could become conscious (washingtonpost.com)

• Plagiarism and Defamation—Two More Bad Things LLMs Are Good At (mindmatters.ai)

🌍 Society

• Sam Altman, OpenAI Movie Filmed in San Francisco Finds a New Home After Amazon Drop (kqed.org)

• Mythos AI: Mythos and the fight over America’s most powerful AI (economictimes.indiatimes.com)

Share

Good morning!

🚀 Launches

• Anthropic finally brings back Claude Fable 5, but you’ll have to live with a temporary usage limit (digitaltrends.com)

• Google releases Nano Banana 2 Lite, its fastest and cheapest AI image generator yet (thenextweb.com)

• OpenClaw is finally available on Android and iOS (techcrunch.com)

💡 Use Cases

• The Best Lifestyle Advice AI Ever Gave Me: 6 Leaders Share What Worked (community.thriveglobal.com)

• I used GPT-5.5 to figure out why my laptop battery was draining, and it found the bug in minutes (xda-developers.com)

⚖️ Policy

• On AI’s 70th Birthday, A Warning About Excessive Regulation (realclearmarkets.com)

💼 Business

• The DeepMind trio who built a poker AI are now making money for quant hedge funds (techcrunch.com)

• This ChatGPT-Powered Stock Picker Helps Busy Entrepreneurs Invest Smarter (entrepreneur.com)

⚠️ Risks

• Top AI Researchers Terrified of a “Chernobyl Moment”: a Mass Casualty Event, or Worse, That Turns the World Against AI Forever (futurism.com)

🌍 Society

• The AI Problem Nobody Saw Coming: The Decline Of Curiosity And Meaning (forbes.com)

Share

1. What is vibe coding? (thehindu.com)

2. Peeter P. Mõtsküla: AI agents need collars, not passports (news.err.ee)

3. Sycophantic chatbots and the harms that build over many chats (helpnetsecurity.com)

4. Chinese startup claims it can match Mythos level performance with new AI model (livemint.com)

5. AI Gatekeepers (blogs.law.ox.ac.uk)

6. ChatGPT Towers Over Claude With 900 Million Users (benzinga.com)

7. Alternatives to OpenAI, Anthropic: With US prime AI off the table, India opts for fine China (economictimes.indiatimes.com)

8. Your Degree Gets You the Job; These AI-Era Skills Could Get You the Pay Rise (ibtimes.co.uk)

9. ‘Drugs, sex and violence: The shocking content AI chatbots are showing children (nzherald.co.nz)

10. I stopped babysitting Claude Code by giving it one persistent goal instead of step-by-step prompts (xda-developers.com)

Subscribe now

Share

The AI story this week begins with an unsettling thought: the biggest change isn’t that AI is becoming smarter, it’s that we’re slowly forgetting where it ends and everyday life begins (economictimes.indiatimes.com).

That shift is happening faster than most people realize. The next generation of AI won’t simply answer questions; it will click buttons, use software, and complete tasks on our behalf, turning computers into something we manage less and delegate more. (forbes.com)

And if the latest industry moves are any indication, those assistants won’t stay trapped inside our laptops for long. The race is already moving toward AI-powered wearables that could follow us everywhere, making the assistant less of an app and more of a permanent companion. (digitaltrends.com)

Investors certainly believe that’s where we’re headed (fool.com). Capital continues to flood into AI companies, driven by the conviction that today’s leaders are building the infrastructure of tomorrow’s economy.

But every technological revolution creates new concentrations of power, and AI is proving no exception (thestatesman.com). As wealth, talent, and computing resources gather around a handful of giants, critics are asking whether we’re witnessing innovation—or the rise of a new digital plutocracy.

That concentration of power depends on something else: data (gulfnews.com). Yet publishers and creators are increasingly refusing to accept the idea that everything published on the web is automatically free fuel for AI models, setting the stage for a defining legal battle over who owns the knowledge that machines learn from.

Meanwhile, the technology itself is revealing a more complicated personality (deseret.com). In one setting, AI is helping lonely seniors reconnect with the world and reducing isolation in meaningful ways.

In another, the same technology is becoming deeply embedded in the lives of teenagers (buzzfeed.com), raising uncomfortable questions about emotional dependence, manipulation, and what happens when an artificial companion becomes more influential than a real one.

Psychologists (psychologytoday.com) increasingly see this not as a test of artificial intelligence, but as a test of human intelligence, of our resilience, our judgment, and our ability to remain emotionally grounded while living alongside machines designed to understand us.

Which is why perhaps the most important lesson this week isn’t about AI at all (forbes.com). As machines become better at logic, speed, and automation, the qualities growing most valuable are the ones they still struggle to reproduce: empathy, communication, trust, creativity, and wisdom.

The AI story is no longer about whether the technology works—it clearly does. The real story is that AI is quietly becoming part of the fabric of everyday life. The next chapter won’t be written by faster models alone, but by how well we decide to live with them.

Subscribe now

Share

1. US lifts Anthropic ban, allows Mythos 5 access for over 100 trusted organisations (nairametrics.com)

2. OpenAI releases GPT-5.6 Sol to 20 government-approved partners in restricted preview (thenextweb.com)

3. AI Tool of the Week: a smart assistant that automates your weekly workflows (livemint.com)

4. How to Disable AI Features on Your Google Pixel Phone (cnet.com)

5. You will never own the AI. You can still own what it cannot. (calcalistech.com)

6. I turned my self-hosted LLM from a glorified chat box into a real AI assistant (xda-developers.com)

7. I ran my local LLM for hours and watched it get dumber in real time (xda-developers.com)

8. The new politics of frontier AI (dailysabah.com)

9. Gemini for Home is a subscription trap disguised as smarter automation (xda-developers.com)

10. I’d do these 5 things differently if I started self-hosting LLMs today (xda-developers.com)

Subscribe now

Share

1. The Trump administration doesn’t want you to use OpenAI’s GPT-5.6 without its approval (neowin.net)

2. Advanced AI could give malicious actors an edge, Cambridge report warns (trtworld.com)

3. AI Boom Drives Global Memory-Chip Shortage, Making MacBooks and Other Gadgets More Expensive (outlookindia.com)

4. AI can speed up defence decisions, but humans must stay in control (straitstimes.com)

5. Companies bought the AI. Now they need people to use it (tech.eu)

6. Edge AI and the Shift from Pilot to Production (eeworldonline.com)

7. CFC introducing affirmative AI coverage (reinsurancene.ws)

8. Anthropic Salaries Revealed: How Much Technical Staff Make in 2026 (businessinsider.com)

9. AI Is Writing Almost All Startup Code. That’s Creating a New Problem. (businessinsider.com)

10. The dark side of AI success: What your employees know that the board doesn’t (cio.com)

Share

A brief note before we begin.

This is still The Brunch News you’ve been reading for the last six months: a curated look at the strange, important, and sometimes alarming things happening around us.

From today on, the focus becomes narrower. I hope that’s good news for everyone.

Over time, the numbers made one thing clear: the stories that drew the most interest were about AI, Big Data, surveillance, privacy, security, algorithms, and the quiet digital machinery shaping our lives. Most of all, they were about digital independence.

So this newsletter is going full speed uncensored.

The Brunch News is becoming The Uncensored GGUF.

The same Brunch News, but now focused on the digital perils of our age. Fewer distant headlines that people politely ignore; more of the technological weirdness already knocking at the door.

Now, let’s talk AI.

1. Pinterest unveils new AI tools for marketers (campaignme.com)

2. Google to save more of your search data for AI training: Here’s how to opt out (indianexpress.com)

3. 60% of Indian consumers interested in creating personal AI agents (aninews.in)

4. ChatGPT and Gemini could be quietly affecting your voting decisions (digitaltrends.com)

5. New Illinois law expands cyberbullying definition to include AI-generated images (kfvs12.com)

6. Anthropic accuses Alibaba of using 25,000 fake accounts to copy Claude’s capabilities (neowin.net)

7. OpenAI is making GPT-5.5 Instant ‘more fun’ to talk to (androidauthority.com)

8. AI is pumping out books. Are they any good? (npr.org)

9. AI Sovereignty and the Implications of the US Export Restrictions (ibtimes.com)

10. AI boom to lift economy and online verification plan’s gift to cybercriminals (irishtimes.com)

11. AI security hackathon identifies weaknesses (ukauthority.com)

12. Convert Ideas to Presentations in Minutes Using NotebookLM (voiceofemirates.com)

Leave a comment