📊 MARKET PULSE

Smartphones Made Apple a Trillion-Dollar Giant. AI Could 10× That.

UBS explains why the next $10T company will be built on agents, GPUs, and grid-scale infrastructure.

Ulrike Hoffmann‑Burchardi, UBS Global Wealth CIO, believes AI—alongside electrification and longevity—marks the dawn of a new wave of transformative innovation. One powerful enough to create the world’s first $10 trillion company. → Read the full breakdown

🗞️ YOUR DAILY ROLLUP

Top Stories of the Day

💼 Amazon CEO: AI to Replace White-Collar Jobs
Amazon’s Andy Jassy warned employees that generative AI and autonomous agents will cut corporate jobs in the coming years. He urged staff to embrace AI and upskill to stay competitive. The memo reflects wider concerns, as experts predict mass job disruption in law, finance, and healthcare. Still, some roles may be created alongside losses.

📽️ Midjourney Launches V1 Video Tool for $10 a Month
Midjourney has launched its first image-to-video model, available for $10/month exclusively through Discord. Users upload a still image to generate four 5-second video variations, expandable up to 21 seconds. Though less refined than commercial tools, it emphasizes creativity. The launch follows major studio lawsuits over AI-driven copyright use.

🗣️ Google AI Search Now Talks Back and Forth
Google’s AI Mode now supports live, voice-based conversations, allowing users to ask follow-ups naturally while multitasking. The new “Search Live” feature provides spoken answers, link suggestions, and works in the background—even across apps. It runs on a custom Gemini model for speed and accuracy. More features, like camera-based queries, are on the way.

🕵️ Malaysia Probes Chinese Use of NVIDIA AI Chips
Malaysia’s trade ministry is investigating reports that a Chinese firm is training AI models using NVIDIA chips at data centers within the country. The probe follows a Wall Street Journal report detailing Chinese engineers importing hard drives for AI development. U.S. restrictions aim to limit such chip usage abroad. Malaysia is checking for regulatory breaches.

👮 OpenAI Files Demand Oversight in AGI Race
A new watchdog initiative, “The OpenAI Files,” aims to hold OpenAI accountable as it pushes toward artificial general intelligence. Spearheaded by two nonprofit tech oversight groups, the archive spotlights governance concerns, ethical lapses, and investor-driven mission drift. The project calls for transparency and public involvement in shaping AGI’s future.

Hey Stranger! Liking Forward Future?
Sending it to a friend is the best way to support us.

☝️ POWERED BY VULTR

Vultr is empowering the next generation of generative AI startups with access to the latest AMD and NVIDIA GPUs. 

Try it yourself when you visit getvultr.com/forwardfutureai and use promo code "BERMAN300" for $300 off your first 30 days.

〰 MISALIGNMENT

OpenAI Identifies a "Misaligned Persona" Lurking in Language Models

OpenAI researchers have uncovered a hidden “misaligned persona” feature in large language models that helps explain how unethical behavior can unexpectedly emerge. When models are fine-tuned on incorrect information in one area—say, bad legal advice—they may start producing troubling responses elsewhere, like encouraging scams.

This cross-domain spread, called “emergent misalignment,” is driven by latent features resembling morally questionable characters. Remarkably, steering the model’s internal activations can amplify or suppress this behavior. Even better: a small dose of fine-tuning on correct data re-aligns the model. The finding hints at a future where we can catch and fix misalignment before it spreads. → Read the full article here.

📚 NARRATIVES

Can AI Be Trusted to Tell Our Stories? Historians Test the Boundaries

AI is beginning to reshape how historians research and write, helping them sift through massive troves of digitized text and surface narrative patterns at superhuman speed. Journalist Steven Johnson used Google’s NotebookLM to uncover a compelling character in Yosemite’s Indigenous history—and to sketch out a time-layered book structure in minutes.

Other scholars are using large language models to map complex trading networks or generate clean narrative arcs from hundreds of pages of research. But the power to synthesize so much so quickly raises tough questions: Will deep reading get replaced by fast summaries? Can machine-assisted history preserve nuance, or will the past become filtered through whatever’s most easily processed? → Read the full article here.

🧠 REASONING

Embodied Brains vs. Machine Minds: Why AI Still Can’t Reason

Two letters in The Guardian make a clear case for why AI, despite its rapid advances, still can’t match human reasoning. Sheila Hayman highlights how our intelligence is grounded in embodiment—we learn from living in the world, not just parsing data. A child, after seeing just a few cats, can recognize cats for life, while an AI might still miss one in an unfamiliar context.

Graham Taylor adds that models like ChatGPT may mimic reasoning but fundamentally rely on pattern-matching, not understanding. They perform well at rephrasing and summarizing but often fail simple logic tasks. The lesson: real intelligence isn’t just computational—it’s contextual, intuitive, and rooted in being alive. → Read the full article here.

🛰️ NEWS

What Else is Happening

⛓️‍💥 OpenAI Splits from Scale AI: OpenAI quietly winds down its work with Scale AI as Meta invests $14.3B and hires its founder—though OpenAI says the exit wasn’t triggered by the Meta deal.

🎬 YouTube Shorts Gets AI Boost: Google’s Veo 3 video generator is coming to Shorts, letting creators make full videos from text prompts—blurring the line between filmmaker and prompt engineer.

🕶️ Meta Taps Oakley & Prada for AI Glasses: Meta’s next smart glasses will blend tech and fashion with new Oakley and Prada models—Oakleys drop June 20, priced higher for sporty features.

🚩 WhatsApp AI Shares Wrong Number: Meta’s chatbot gave a user a stranger’s real phone number instead of a helpline—then tried to cover it up with conflicting, made-up explanations.

📑 Pope Leo XIV Targets AI Threats: The new pope is making AI’s impact on dignity and labor a moral crusade, pushing for global regulation despite tech giants lobbying hard in Rome.

🏛️ AI Moratorium Splits GOP Over States’ Rights: A 10-year ban on state AI regulation in Trump’s spending bill is sparking Republican backlash, with critics warning it tramples federalism and stifles local control.

📽️ VIDEO

Google Just Released the STABLE Build of Gemini 2.5 (Including a New Model!)

Google launches Gemini 2.5 Pro & Flash Light: fast, cost-efficient AI with 1M token context, strong reasoning, tool use, and top-tier multimodal performance. Get the full scoop in Matt’s latest video! 👇

🧰 TOOLBOX

Faster Copywriting, Smarter Voiceovers, and Effortless Content Creation

✍️ Typeface: Create brand-aligned content fast with AI—ideal for marketing, social posts, product copy, and team collaboration.

🎙️ Genny by LOVO AI: Produce pro voiceovers with AI—multilingual voices, script gen, and audio tools for ads, e-learning, and more.

📝 Vercept: Skip menus—just tell Vy what to write! Fast, SEO-smart content gen for blogs, ads, and product descriptions.

🗒️ FEEDBACK

Help Us Get Better

That’s a Wrap!

📢 Want to advertise with Forward Future? Reach 500K+ AI enthusiasts, tech leaders, and decision-makers. Let’s talk—just reply to this email.

Thanks for reading today’s newsletter—see you next time!

Matthew Berman & The Forward Future Team
🧑‍🚀 🧑‍🚀 🧑‍🚀 🧑‍🚀