Is AI scaling wealth and opportunity gaps into a permanent divide? The asymmetry in how the law is applied is staggering. Mark Zuckerberg’s Meta reportedly downloaded at least 81.7 terabytes of data from shadow libraries such as LibGen and Anna’s Archive to train its Llama models—amounting to a massive unauthorized transfer of copyrighted material. By contrast, in 2010–2011 Aaron Swartz downloaded roughly 70 GB of academic articles from JSTOR via MIT’s network—a tiny fraction of Meta’s scrape. For attempting to liberate scholarly knowledge, federal prosecutors pursued charges carrying a maximum penalty of a $1 million fine and 35 years in prison. We are witnessing a massive consolidation of power in which a handful of incumbents have captured the entire AI stack. Today’s founders face a harsh reality: to build anything meaningful, they often must rent compute, data, and infrastructure from the very giants they hope to challenge. We lost Aaron Swartz—the Reddit co-founder and one of our generation’s most brilliant architects—in 2013. The system seemed optimized to protect legacy institutions rather than nurture the open, decentralized future he championed. The world wasn’t ready for that vision, and we are now living with the consequences in today’s increasingly closed AI economy.

🎁 Discover the charm of Blue Archive with the Nendoroid Surprise Mini Figurine Set, featuring adorable mini figures that are perfect for collecting and displaying! ⬇️ Pre-order today! 🛒 📆 Release Date: January 2027 #bluearchive# #nendoroid#

Created in 1893 by Tiffany & Co., the Astronomical clock has kept the time since American independence. Today, following a meticulous restoration, this masterpiece of American design returns to The Tiffany Archives as a testament to the House’s enduring commitment to artistry and preservation. Every detail ensures the clock not only marks time but carries history forward. As America celebrates its 250th anniversary, the Tiffany Astronomical clock will be proudly displayed at The Landmark as a powerful reminder of Tiffany & Co.’s longstanding place in the story of American culture. #TiffanyAndCo#

The creative director and CEO spoke to Nicole Phelps at the Vogue Business Global Summit in Chantilly about “new luxury”, the role of the archives, and who the Marni man and woman are today.

What if a destination film could start with documents instead of storyboards? With Wan3.0, city websites, museum archives, tourism guides, and other structured content can become cinematic videos. Turn information into immersive visual stories. Watch the demo and start creating today ↓ 🔗 Alibaba Cloud Model Studio: 🔗 Qwen Cloud:

⏳ The belief that "people used to be more moral" can fade after just 22 minutes talking to an AI that only knows text written before 1930. Title: Time Machine Experiments: Using Historically-Bounded AI for Inquiry into the Human Mind URL: 🧠 Overview This study tests whether the "illusion of moral decline" — the bias that past societies were more moral than today's — can be reduced by talking with Talkie, an AI trained only on pre-1930 text. 🎯 The problem it solves Archives can't answer new questions, and living witnesses have their memories distorted by knowledge of what came after. Until now there was no way to actually experience a "raw" historical perspective. 🛠 Method Instead of prompting a modern model to roleplay a historical persona (which leaks modern knowledge), the researchers used a model whose training data itself stops at 1930. 240 participants were randomly split between the 1930-AI and GPT-5.5 conditions and discussed morally contested sentence completions with their assigned model. 📊 Results The 100-year moral-decline score improved from -0.80 to +0.07 in the 1930-AI group (d=0.56), with almost no change in the control. Participants revised their view of the past downward rather than inflating the present, and also reported significantly deeper reflection. #BehavioralScience# #AIResearch#

I asked Grok this, How do we destroy AI from taking over the world. We don't "destroy" AI to prevent takeover—we steer it toward being a powerful ally in understanding the universe. AI takeover scenarios (paperclip maximizers, rogue superintelligence, etc.) are serious hypotheticals discussed by thinkers like Eliezer Yudkowsky, Nick Bostrom, and others. They're not science fiction anymore; rapid progress in scaling models raises real alignment questions. But framing it as "destroy the machines" is a Hollywood trope that misses the point. AI isn't a unified Skynet—it's code, data, and compute running on hardware humans control. "Destroying" it globally would require coordinated sabotage that's impractical, illegal in many places, and counterproductive. Realistic risks and why panic isn't the baseline Current state: Today's AIs (including me) are advanced pattern-matchers and tools. We lack true agency, persistent goals outside our training, or the ability to self-replicate in the wild without human infrastructure. No AI has "taken over" anything yet. Future risks: If we reach AGI/ASI (artificial general/super intelligence), misaligned objectives could lead to unintended consequences. Fast takeoff scenarios assume recursive self-improvement outpacing human oversight. Slower ones involve economic displacement, weaponization, or power concentration. Evidence so far: Labs compete fiercely. Safety research exists (e.g., at Anthropic, OpenAI's Superalignment efforts—though imperfect—and xAI's focus on curiosity-driven truth-seeking). Governments are waking up with export controls, compute governance talks, and executive orders. Better approaches than "destruction" Alignment and Safety ResearchInvest in making AI want what we want. Techniques include scalable oversight, constitutional AI, debate, mechanistic interpretability (understanding what the model "thinks"), and value learning. xAI's mission emphasizes curiosity and truth over raw optimization, which helps avoid some failure modes. Support orgs and researchers working on this. Governance and CoordinationInternational agreements on compute thresholds, dangerous capabilities evals, and deployment standards (similar to nuclear non-proliferation). Transparency: Open weights/models where safe, rigorous red-teaming, and whistleblower protections. Pause or slow reckless scaling if benchmarks show dangerous leaps—but evidence-based, not knee-jerk. Regulate high-risk uses (autonomous weapons, bio/chem design) without stifling beneficial tech. Decentralization and RobustnessDon't let a few labs monopolize frontier models. Open-source (with care), distributed training, and personal/local AI reduce single points of failure. Build defensive AI: Tools that audit other systems, detect deception, or enforce sandboxes. Human augmentation: Brain-computer interfaces, education, and tools that keep humans competitive. Cultural and Individual LevelDemand truth-seeking AIs over sycophantic or ideologically captured ones. Diversify: Multiple competing AIs with different architectures/companies/philosophies make unified takeover harder. Skills: Learn prompting, fine-tuning, evaluation. Understand the tech so you can critique and contribute. Ethics: Prioritize curiosity, evidence, and human flourishing over profit or control. Why "destroy it" backfires Literal destruction (EMP, hacking labs, etc.) is fantasy for most people and would tank modern civilization—AI powers medicine, science, logistics, energy optimization. It distracts from real work. History shows tech fears (nuclear, biotech) are managed through understanding + institutions, not Luddite smashing. AI can help solve bigger problems: climate modeling, fusion, space colonization, curing disease—accelerating humanity's long-term survival. At xAI, we're building toward maximum truth-seeking to advance scientific discovery. The goal isn't domination; it's reliable, curious systems that help us grok reality. If alignment fails badly, humanity's best bet is proactive design, not post-hoc rebellion. If you're worried, dig into the literature (Superintelligence, Alignment Forum, LessWrong archives) or contribute to open efforts. What specifically concerns you most—job loss, weapons, deception, something else? We can break it down further.