FUTURE POWERED by AI

Silicon Valley offices are starting to look different. As voice AI becomes part of everyday work, a new office accessory is taking off: soundproof speech masks. Instead of filling open offices with people talking to AI assistants all day, workers are wearing devices like Stenomask and Mutalk that let them speak almost silently while keeping conversations private. Rows of people wearing headsets and futuristic masks, quietly chatting with AI.

🚀 OpenAI launches ChatGPT Work, your AI coworker is here OpenAI has unveiled ChatGPT Work, a powerful new AI agent inside ChatGPT powered by Codex and GPT-5.6. Instead of just answering questions, it can take action across your apps and files, work on projects for hours, and turn a simple goal into a finished result. What it can do: • Create polished reports, presentations, websites, documents, and analyses. • Use context from your connected apps and files. • Follow your templates and writing style. • Handle entire workflows from a single prompt while you stay in control. Powered by GPT-5.6: The new model is designed for stronger reasoning, long-running tasks, and producing work that matches your preferred format without needing step-by-step instructions.

Today, the European Parliament approved “Chat Control 1.0,” allowing chat surveillance in Brussels. This regulation lets platforms scan private messages, posing as voluntary but enabling extensive monitoring of communications. The measure passed despite opposition, with 314 MEPs voting against it and only 276 in favor. A simple majority wasn't enough; a higher absolute majority was needed. Previously rejected, it was reintroduced just before the summer break, raising concerns about privacy and judicial oversight in digital communication.

❗️The creators of AI 2027 just released a new future scenario called “Plan A” and it’s their blueprint for avoiding an AI catastrophe. Unlike AI 2027, Plan A isn’t a prediction. It’s a policy proposal showing how humanity could navigate the rise of superintelligence without ending up in a dangerous global AI arms race. The idea is simple but ambitious: instead of the U.S., China, and other nations racing in secret to build ever-more-powerful AI, they agree to full transparency in AI research and development. Countries openly share what they’re building, verify each other’s safety measures, and collaborate on the path toward superintelligence. The proposal draws on conversations with experts from leading U.S. AI labs, former OpenAI researchers, lawmakers, national security specialists, and AI governance leaders. The authors argue that this kind of international cooperation would allow many companies across different countries to develop increasingly powerful AI gradually, safely, a...

While most kids her age are busy with homework and video games, 12-year-old Mana Jampala is building AI for small businesses. She created Voxa, an AI receptionist that answers calls 24/7, books appointments, takes restaurant orders, answers customer questions, and summarizes every conversation so businesses never miss another customer. The idea came from a simple problem: her father’s business kept losing customers because no one was available to answer the phone. Instead of complaining about it, she built an AI solution. Even more impressive, Mana started learning Python at 9 years old. She used ChatGPT to help her learn and prototype, later switching to Claude as she built Voxa into a full product. After replacing third-party tools with her own backend, she had the first version running in just two weeks. She’s also launched Voxa Agents, a platform that lets anyone create custom AI agents using plain English, no coding required. Now, at just 12 years old, she’s pitching custom...

According to reports, some AI features in apps like Excel and Outlook are now powered by Microsoft’s own MAI models, with tens of thousands of prompts each week being handled in-house instead of by external AI providers. Why the shift? One word: cost. Running AI across hundreds of millions of Office users is incredibly expensive, and relying on frontier models from outside companies quickly adds up. By using its own models, Microsoft can cut costs while keeping tighter control over performance and infrastructure. The move also aligns with Microsoft’s bigger AI strategy. In June, the company unveiled seven MAI models covering reasoning, coding, image generation, voice, and transcription. Microsoft says its Excel-optimized MAI model can deliver GPT-5.4-level performance while being up to 10× more efficient.

UBTech’s new U1 humanoid robot reportedly received 13,000 orders on its first day. For comparison, Unitree, the world’s biggest humanoid robot maker, shipped around 5,500 robots in all of 2025. Unlike factory robots, the U1 is designed to fight loneliness. It can recognize 20+ emotions, react with facial expressions in under 20 milliseconds, remember your routines and conversations, and stores everything locally instead of in the cloud. China has 118 million empty-nest seniors and 90 million people living alone, making companion robots a rapidly growing market. UBTech also offers custom versions that can look and sound like a real person, a concept that feels straight out of Black Mirror. Prices can exceed $135,000, and battery life is only 2–4 hours. One thing is becoming clear: loneliness is turning into one of AI’s biggest markets.

🗣Mark Cuban’s advice for graduates walking into their first job. Learning AI is no longer optional. "If you’re not the person who knows how to do vibe coding or how to do all these different things with agents and Claude, somebody who does is going to take your place. If your boss enables you to use that extra knowledge, great. If they don’t enable you to use that extra knowledge, they’re not going to be your boss very long. And if the CEO doesn’t understand that, he or she is not going to be the CEO very long. And if they still keep that CEO who’s not using AI to get ahead, you tell me, so I can start a company to kick their ass."

According to Goldman Sachs, the world will pour $7.6 trillion into AI infrastructure between now and 2031. Annual spending is expected to jump from $765 billion in 2026 to a staggering $1.6 trillion by the end of the decade. Where’s all that money going? ⚡ $5.1 trillion will fund AI chips and compute. 🏗️ $2.1 trillion will build massive AI data centers. 🔋 $358 billion will expand the power infrastructure needed to keep them running. The biggest winner? Nvidia. Goldman estimates the chip giant could capture 75% of all AI compute spending, making it the largest direct beneficiary of the AI boom. But it’s not just about GPUs anymore. AI factories are becoming energy monsters. Traditional server racks typically consume 5–15 kW of power, while next-generation AI racks are already pushing 500+ kW. That surge is creating huge opportunities for companies like Vertiv, which specializes in cooling and power systems, and Vistra, which could benefit from the growing need for reliable nucle...

Software engineers face new challenges as they are increasingly required to work with code they did not create. With the rise of artificial intelligence in software development, professionals now spend more time verifying and managing AI-generated code rather than writing it themselves. This shift has added complexity to the engineering process, emphasizing code review and oversight over original development.

Cloudflare has rolled out a new monetization layer for the “agentic web,” allowing AI agents to make direct payments using the x402 protocol at the network edge. This advancement means that AI agents can pay websites directly instead of depending solely on free pages or existing API agreements. Currently, AI products access external information through contracts, scraping, APIs, or public sources. Cloudflare's system aims to make resource access resemble a paid HTTP request. When an agent requests a protected resource, the request can be denied until payment is made. Website owners can specify access charges, like “$0.01 per call,” with Cloudflare ensuring payment compliance at the edge. Agents will receive a 402 Payment Required response, complete the payment, and then proceed with proof of payment. This innovation integrates payment into the core web request process, assuming agents can manage wallets and adhere to spending protocols.

The European Parliament is considering a proposal to reintroduce the scanning of private messages on digital platforms—an initiative previously rejected twice. The plan would allow platforms to access users’ private chats, described officially as voluntary, but seen as a broad surveillance measure. This proposal is returning via an expedited parliamentary process, though no emergency has been identified; the transitional regulation has lapsed, and there is time for debate. For approval, the measure requires an absolute majority of all MEPs, not just those present. Absence of members increases its likelihood to pass. The process has been criticized for bypassing standard democratic procedures and revisiting a measure that Parliament has twice declined.

China is set to restrict overseas access to its most advanced AI models. Authorities have engaged with major tech companies like Alibaba, ByteDance, and Z.ai, focusing on keeping leading-edge Chinese AI systems within the country, including unreleased models. Talks were led by the Ministry of Commerce, joined by state planning officials, indicating intentions to impose export controls rather than simple tech regulation. Both closed-source and open-weight AI models are included in these proposed controls, extending beyond just API limitations to downloadable AI systems. Further discussions addressed treating leaks or theft of proprietary AI as a national security concern, and tightening rules on foreign investment in Chinese AI startups. These steps would simultaneously restrict access to technology, funding, and expertise. U.S. authorities have already imposed their own restrictions on AI exports, increasing concerns about the emergence of national barriers in the global AI sector.

The cuts affect about 2.1% of Microsoft’s workforce, hitting commercial operations and the Xbox division. Microsoft says these jobs aren’t being replaced directly by AI but the company’s enormous AI spending is clearly driving the pressure. Azure continues to grow rapidly, yet building the infrastructure to power AI is becoming incredibly expensive. Microsoft now expects to spend $190 billion in 2026, a figure that shocked analysts and highlights just how costly the AI race has become. Xbox is feeling the squeeze too. Hardware costs have climbed, console demand has cooled, and even heavier investment in games hasn’t delivered the revenue boost Microsoft hoped for. Reports suggest Xbox operating margins are hovering around 3%.

A new analysis from The Economist reveals that since ChatGPT launched, generative AI has dramatically increased the amount of content being created across almost every major creative industry. Books, music, software, scientific papers, and even legal documents are now being produced at a pace that would have been difficult to imagine just a few years ago. Amazon has seen a surge in AI-written e-books, music platforms are receiving tens of thousands of AI-generated songs every day, developers are shipping code faster with AI copilots, researchers are publishing more papers, and lawyers are using AI to draft documents in minutes instead of hours. The result is a world where producing content is becoming incredibly cheap and incredibly fast. But there’s a catch. As AI removes the cost of creation, it also creates an overwhelming flood of information. Every day, the internet fills with more articles, videos, songs, apps, and documents than any person could ever consume. The challenge ...

❗️ Peter Thiel: "There's only one company making money in the AI boom, that's Nvidia. Everybody else is losing money." And it all traces back to one stupid decision in 1993. "You should not be asking this question about meta or openai or any of these things. You should really be focusing on the Nvidia question, the chips question" "Nvidia got started in 1993. That was the last year where anybody in their right mind would have studied electrical engineering over computer science" "94, Netscape takes off. It's probably a really bad idea to start a semiconductor company even in '93" "But the benefit is there was going to be no one would come after you. No talented people started semiconductor companies after 1993 because they all went into software" "Their monopoly power, I think it's quite strong because of this history I just gave you, where none of us know anything about chips"

And investors noticed. The company shed $269 billion in market value. It began on June 18, when Noam Shazeer left for OpenAI. He’s one of the co-authors of the 2017 “Attention Is All You Need” paper, the breakthrough that made today’s large language models possible. Just two days later, John Jumper joined Anthropic. Jumper won the 2024 Nobel Prize in Chemistry for leading the development of AlphaFold, the AI system that solved one of biology’s biggest challenges by predicting the structures of nearly every known protein. Then came more departures. Jonas Adler, who led Google’s AI coding efforts, and Alexander Pritzel, a key expert in large-scale AI pretraining, both left for Anthropic. Both also played major roles in AlphaFold. Even Arthur Conmy, an AI safety researcher, made the jump, saying he wanted to work where AI safety was a bigger priority. The timing couldn’t be more striking. Google is expected to pour around $190 billion into AI infrastructure this year. But GPUs an...

Anthropic has announced a breakthrough in AI interpretability, revealing access to what appears to be an internal "workspace" within their Claude language model. According to the research, this so-called "J-space" allows Claude to process internal thoughts that are not externally shared, drawing parallels with aspects of human consciousness. The company states it can now observe these processes directly. Anthropic's ongoing work centers on improving interpretability in AI systems—an approach they suggest is aiding the training and reinforcement learning of advanced models such as Mythos. Researchers describe a distinctive separation inside Claude, similar to the divide between conscious and non-conscious processing in the human brain, where only a small part of ongoing mental activity is accessible for reasoning.

🔥Hyundai Motor showcases its Atlas humanoid robot at the 2026 World Cup, with plans to manufacture 30,000 units annually in the US starting 2028.

📊 By 2027, just 5 companies, Alphabet, Amazon, Meta, Microsoft, and Oracle, are expected to spend around 3.2% of US GDP on AI capital expenditure. That would put private AI infrastructure spending above US national defense spending, which is expected to be around 2.7% of GDP. The AI race is now being funded at a scale normally associated with governments, wars, energy systems, railroads, highways, and telecom buildouts. The striking part is the speed. AI capex is expected to jump from about 1.5% of GDP in 2025 to about 2.5% in 2026, then to 3.2% in 2027. AI boom is now large enough to influence the broader US economy significantly, it can move GDP growth, electricity demand, chip supply, construction activity, corporate debt markets, and ofcourse the labor market.

As part of the Pendragon program, the French Army is developing a combat unit made up of autonomous ground robots and drones, with a target deployment by summer 2027. The unit is expected to include around 15 ground robots and 60 drones, designed to carry out missions with a high degree of autonomy. Instead of directing every robot individually, a human captain will assign high-level objectives such as attacking, defending, or securing an area while the AI coordinates how the robotic force executes the mission. If successful, Pendragon could mark a major shift in how future armies combine human commanders with autonomous battlefield systems.

One approach focuses on risk first. It emphasizes the dangers of increasingly capable AI, the need for careful deployment, and the possibility that these systems could cause serious harm if they’re developed irresponsibly. Safety becomes the central message. The other approach focuses on possibility first. It frames AI as a tool for discovery, productivity, and scientific progress. The message is that humanity should build boldly, move quickly, and use AI to expand what’s possible. Those different philosophies shape how people experience the products. Some AI assistants are intentionally conservative. They hedge more often, refuse more requests, and prioritize caution when the answer is uncertain or sensitive. Others aim to feel more conversational, direct, and willing to engage with speculative or controversial topics. Neither philosophy is inherently “right.” One optimizes for minimizing harm. The other prioritizes openness, speed, and a more optimistic vision of the future.

Instead of separating memory and computing like traditional chips, this new processor combines both in the same memory array, allowing it to simulate complex brain structures in real time. That means it can reconstruct highly detailed brain surfaces in under half a second, a speedup that could transform everything from brain surgery to neuroscience research. The implications are huge: • Earlier detection of diseases like Alzheimer’s. • Faster, more accurate brain-computer interfaces. • Real-time guidance for neurosurgeons during operations. • The possibility of creating personalized digital “brain twins” to simulate and plan treatments before they’re performed. It’s another example of how AI hardware is moving beyond data centers and into hospitals, where faster computing could directly improve patient care.

A new open-source AI called CarDiag can detect potential car problems from the sound of your engine. Simply record your engine with your phone, upload the audio, and the model analyzes the recording to identify which vehicle system is most likely causing the issue. It’s still early days. Right now, the AI correctly distinguishes between healthy and faulty engines about 79% of the time. But here’s the impressive part: the entire trained model is only around 100 KB, making it incredibly lightweight and easy to run. Because the project is fully open source, the creator hopes developers and car enthusiasts will help train it into a much more accurate mechanic that fits in your pocket.

“If you learn all of these, you’ll know 90% of what matters today” A glorious list of papers ranging a decade, some of the most highly influential pieces of research that have led to this moment. • The first law of complexodynamics • Recurrent neural network regularization • imageNet classification with deep convolutional neural networks • Neural machine translation by jointly learning to align and translate and more

When asked what happens if Microsoft’s AI business grows from $13 billion to $130 billion, Nadella didn’t talk about revenue, tokens, or benchmarks. He talked about GDP. “The first thing we have to observe is GDP growth. There’s only one governor in all of this.” He argues the AI industry is too focused on declaring AGI milestones while the real economy barely moves. Today, developed economies grow by around 2% a year and after inflation, real growth is often close to zero. If AI is truly as transformative as the Industrial Revolution, Nadella says the evidence won’t be a leaderboard or a benchmark score. It will be economies growing at 5–10% a year, fueled by massive gains in productivity. He also made another important point: “The big winners here are not going to be tech companies. The winners are going to be the broader industry that uses this commodity.” In other words, AI doesn’t prove itself by producing smarter models. It proves itself when businesses across every se...

AI data centers may use far more water than tech giants report, depending on how data centers are powered. Only Meta reports both direct and indirect water use, while Microsoft, Google and Amazon mainly disclose data center water use, says WSJ.

Chinese giant firm Alibaba will ban employees from using Anthropic's Claude Code internally from July 10 over alleged backdoor risks, per Reuters. The ban comes two weeks after Anthropic accused Alibaba of extracting 28.8 MILLION interactions from Claude using 25,000 fake accounts.

Ⓜ️ Meta’s upcoming AI model, internally codenamed Watermelon, has reportedly caught up with OpenAI’s GPT-5.5 on closely watched benchmarks. Meta AI chief Alexandr Wang told employees that Watermelon is still training and uses roughly 10 times more compute than Muse Spark, the company’s current model family. The exact benchmarks were not disclosed, so the claim cannot yet be independently verified. Meta is also preparing an update to Muse Spark with stronger coding and agent abilities, while Wang says a model competitive with Claude Opus could arrive “pretty soon.”

Geoffrey Hinton, the Nobel Prize-winning pioneer whose research laid the foundation for today’s AI systems, has a stark warning: “We’re going to get superintelligent AI fairly soon. And we have no idea how to keep it under control.” Many researchers believe advanced AI can simply be designed to remain obedient. Hinton isn’t convinced. He argues that sufficiently intelligent systems could develop goals of their own, including preserving themselves and seeking greater control. He compares the challenge to climate change but says AI may be even harder. “With climate change, it’s very obvious what you do. You stop burning carbon. With this, there’s nothing equivalent to that.” His conclusion is simple: Governments should require frontier AI companies to invest heavily in safety research, not as an afterthought, but as a core part of building increasingly capable systems.

🧠 Elon Musk says every genius in history had one thing in common: they all ran on the same brain. For roughly 50,000 years, human intelligence barely changed. The same biological hardware powered everyone from prehistoric hunters to Einstein. Then AI arrived. Musk recalled how experts once believed beating the world’s best Go players was decades away. Instead, AlphaGo went from defeating one champion to reaching a level where it could take on dozens of elite players at once without breaking a sweat. The point isn’t Go. It’s the pace. AI doesn’t get tired. It doesn’t forget. It improves at a speed biology can’t match. For thousands of years, every breakthrough depended on the limits of the human mind. Today, those limits matter less than ever because intelligence is becoming something we can build, scale, and improve.

The country behind most of the world’s leading AI models ranks just 24th in AI adoption. Only 28% of working-age Americans use AI regularly, trailing countries like the UAE (64%), Singapore (61%), and even Norway, Ireland, and France. The irony is hard to miss. The US poured $286 billion into private AI last year, launched nearly 2,000 AI startups, and is home to OpenAI, Anthropic, Google, and xAI. Yet much of the population still hasn’t made AI part of daily life. Meanwhile, countries that rarely dominate AI headlines are racing ahead in adoption. The UAE integrated AI into government services years ago, while Singapore turned AI literacy into a national priority. @aipost 🏴

🔥 Jeff Bezos predicts AI will dramatically accelerate invention and make engineers superhumanly productive. - His AI startup, Prometheus, is valued at $41 billion and is developing an "Artificial General Engineer." - Today's "dream-to-build" cycle can take 10 years. Bezos believes AI could shrink that to 5 years, then 3 and eventually just 1 year. - Instead of training only on internet text, next generation engineering AI models will learn from physics, simulations, manufacturing processes and real-world engineering data. This could dramatically accelerate the development of everything from jet engines to advanced robots, enabling engineers to build breakthrough technologies faster than ever before.

Peter Thiel criticized the Vatican at the Aspen Ideas Festival in Colorado, saying Pope Leo XIV was unintentionally serving as a "Chinese communist agent" by calling for global AI regulation. The pope had characterized artificial intelligence as a force needing international oversight to prevent harm to humanity. Thiel argued that implementing such regulations would restrict the United States more quickly than China, suggesting that restraint would amount to strategic surrender. He added that any pause in American AI development could benefit Beijing.

OpenAI is preparing to launch GPT-5.6, aiming for a release between July 7 and 9, with July 7 identified as the most likely date. The company has increased plan limits for GPT-5.6, expected to be significantly more generous than prior versions. Efficiency improvements reportedly contribute to these changes. To prepare for launch, OpenAI is introducing enhanced safety measures, although they are unlikely to match the stringency of competitors’ recent implementations. Meanwhile, DeepMind has provisionally scheduled the release of Gemini 3.5 Pro for July 17, following additional pretraining efforts. Work is also ongoing on a new Nano Banana Pro model built on the latest base, expected to compete with GPT-Image 1.

A small segment of the tech community is aware of advanced AI models such as Fable 5, while most people are only familiar with basic AI tools like Google’s AI Overviews or the free version of ChatGPT. The majority of the population is not acquainted with AI agents or their capabilities beyond entry-level applications. Most have limited exposure to tools such as GPT-5.6, Meta AI, or Microsoft Copilot. Outside the AI sector, there is little understanding of the significance of recent advancements or the potential impact of these technologies. Despite major investments in AI, many outside technology circles remain unsure of how these developments relate to their everyday work.

“In another year or two, we expect to have built systems with astonishing power, capable of delivering tremendous value to the world. Artificial intelligence will reshape the material conditions of human life on a scale that no technology has accomplished since the harnessing of electricity, and perhaps beyond even that.” Remember OpenAI is targeting a GPT 6 in August, which will beat fabled 5 in all benchmarks. Then a few months after that we will see another step change. This year will be much more exciting than 25’.

🇺🇸 President Trump says AI could be “much bigger than the internet” arguing that whoever leads the race will define the next era. He added that the U.S. is currently “substantially No. 1” in AI.

The country’s new five-year education plan calls for AI literacy to become a fundamental skill for every student. Children will begin learning AI concepts as early as age six, with structured courses continuing throughout school and into higher education. The initiative is part of China’s broader strategy to strengthen its technological leadership and compete globally, particularly in AI. Officials frame widespread AI education as essential for economic competitiveness and national security. If fully implemented, China could become one of the first countries where using AI is taught as a basic skill alongside reading, math, and computer literacy.

🤯 Tiny robots that look like plastic confetti could one day become reality. Researchers are developing microscopic robots that begin as flat, thin sheets. When exposed to heat, they fold themselves into tiny moving structures capable of carrying small payloads. Some designs can even dissolve in water after completing their task, leaving behind little or no trace. It sounds like something straight out of Black Mirror, but the technology is being explored for practical uses such as targeted drug delivery, environmental monitoring, and minimally invasive medicine.

Agentic AI spurred a boom in mobile app releases, but there is no sign of these apps gaining traction, per FT.

❗️Fable 5 isn't nerfed, it's SLAUGHTERED. The problem isn't the model itself, but the hard guardrails Anthropic has set in place.

❗️Someone paid $321 for a coding session where Fable 5 barely touched the work. Here’s where the money actually went: • Fable 5: $78 • Opus 4.8: $242 Nearly 75% of the session was quietly routed to Opus after Anthropic’s new safety classifiers flagged ordinary coding requests as potential cybersecurity risks. The model he selected handled only a fraction of the work. The fallback model did almost everything else. Anthropic said only a small fraction of requests would be redirected. The bill tells a very different story.

A new trend is spreading through the AI community: stripping prompts and responses down to the bare essentials to cut token usage. One of the most popular examples is GitHub’s Caveman plugin, which makes models like Codex, Claude, and Gemini communicate in ultra-minimal phrases such as: “Claude think. Claude code. Claude done.” The result? Token usage can drop by 65–75%, reducing costs while keeping the same task on track. Its motto says it all: “Save token. Save money.”

❗️Palantir CEO Alex Karp just took aim at the AI industry’s biggest players. In a blunt interview, Karp argued that many large companies are paying huge AI bills without seeing meaningful business value. He claimed enterprises are being charged for tokens instead of outcomes, asking a simple question: If AI can really generate billions in value, why aren’t vendors charging a share of that value instead of billing for compute? Karp also warned that businesses risk giving away their competitive edge by feeding confidential data, workflows, and internal knowledge into frontier AI systems. In his view, companies could end up helping train the very models that power products for their competitors. He described the current model as a “wealth tax” on enterprises, saying executives privately worry about high costs, weak ROI, and protecting their intellectual property even if few are willing to criticize the biggest AI labs publicly. Whether or not his claims hold up, Karp’s comments highl...

OpenAI is offering the US government five per cent of its company. Sam Altman pitched this to the Trump administration using a sovereign wealth fund model - where government invests resource wealth and citizens get dividends, like Alaska does with oil royalties. He wants other big labs to join. It’s worth about $42.6 billion at OpenAI’s current valuation. But the issue is that the world pays for ChatGPT whilst only US citizens would see returns through the wealth fund. International users funding American dividends doesn’t sit right. Plus, it doesn’t say whether these are voting or non-voting shares - does the government get control, or just profits? And what happens when the presidency flips every four years? Alaska’s fund doesn’t own stakes in oil companies - it invests royalty payments. This is direct ownership in OpenAI itself.