🗞️ YOUR DAILY ROLLUP

Top Stories of the Day

📱 Google’s Pixel and Watch Just Got Smarter With Gemini AI
Google’s Gemini AI now runs locally on Pixel 10 devices, using “Magic Cue” to surface relevant info from Gmail, Calendar, and more when you need it. The Pixel Watch 4 adds Gemini as a personal health and sleep coach. These upgrades push Android devices closer to being true AI assistants—context-aware, fast, and personalized.

☢️ AI Model Boosts Fusion Power Predictions
A new deep learning model from Lawrence Livermore scientists predicted the outcome of a 2022 nuclear fusion experiment with 74% accuracy—more precise than traditional supercomputers. Trained on over 150,000 simulations, it uses Bayesian inference to combine real data with virtual experiments. The model could speed up fusion research.

🖼️ Google Photos Adds AI-Powered “Edit by Asking”
Google Photos now lets users edit images by simply describing changes via text or voice. Debuting on Pixel 10 devices, the feature uses Gemini AI to handle tasks like removing objects, adjusting lighting, or even adding creative elements. It also introduces C2PA Content Credentials to clarify when AI is involved, boosting image transparency.

📄 Adobe Reinvents PDFs with Generative AI Tools
Adobe has transformed its iconic PDF format by embedding generative AI into Acrobat Studio and new “PDF Spaces.” These virtual hubs let users collaborate with AI assistants that can summarize, cite, brainstorm, and organize content. It’s a leap from static forms to interactive digital workspaces. In 2025, even PDFs aren’t immune to AI’s reach.

Enjoying our newsletter? Forward it to a colleague—
it’s one of the best ways 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.

👾 FORWARD FUTURE ORIGINAL

Inside GPT‐5: OpenAI’s Mark Chen on Reasoning, Synthetic Data, and the Future of Intelligence

GPT‑5 is here, and it’s more than just a bigger model. According to Mark Chen, Chief Research Officer at OpenAI, GPT‑5 represents a critical convergence between traditional pre-training and post-training with deep reasoning capabilities. In this in-depth interview, he explains what sets the model apart, why synthetic data is becoming essential, and how OpenAI balances ambition with responsibility.

From personal “vibe check” tests to training decisions and memory architecture, Mark takes us behind the scenes of one of the most anticipated model launches in AI history. → Read the full article here.

📊 MARKET PULSE

Bluefish Scores $20M to Navigate the Wild West of AI Marketing

When Adidas wants to know what ChatGPT is saying about their latest sneaker drop, they don’t just cross their fingers anymore. They turn to Bluefish—fresh off a $20M Series A led by NEA with backing from Salesforce Ventures and others. CEO Alex Sherman calls it “the biggest shift in consumer discovery since Google”: if someone asks AI about your brand, you’d better have a strategy for the answer. → Continue reading here.

🎭 IDENTITY

He Licensed His Face to TikTok. Now His AI Twin Is Selling Supplements—and Regrets

Actor Scott Jacqmein licensed his likeness to TikTok for $750, expecting a modest boost to his career. Instead, his AI-generated avatar now appears across social media promoting insurance, horoscopes, and supplements—sometimes in Spanish, a language he doesn't speak. Like others, he wasn't told his image could be used beyond TikTok, including on ByteDance's other platforms.

TikTok's avatar tool lets advertisers deploy AI versions of real people without paying royalties. For under $1,000, actors gave up long-term rights, often unknowingly. As generative AI floods digital marketing, the real humans behind these avatars are losing control over their identities—sometimes in embarrassing ways. → Read the full article here.

💻 CYBERSECURITY

Hackers Are Using AI to Steal Billions—And Cybersecurity Firms Are Cashing In

AI has turbocharged cybercrime, enabling hackers to create sophisticated malware, deepfakes, and targeted phishing attacks with unprecedented speed and scale. Even basic generative models like ChatGPT can be manipulated to write malicious code, as demonstrated by researchers who built working malware in hours. Tools like XanthoroxAI offer criminal-grade deepfake and social engineering capabilities for just $150 a month.

But while AI has made life easier for cybercriminals, it’s also sparked a boom in cybersecurity. Firms like Palo Alto Networks and Microsoft are doubling down on acquisitions and AI-focused tools, as demand for digital defenses surges. The arms race is officially algorithmic. → Read the full article here.

⚛️ SCIENCE

NASA and IBM Train AI on the Sun to Predict Solar Storms, and Protect Earth’s Tech

NASA has launched Surya, a powerful AI foundation model built with IBM and trained on nine years of solar data to forecast solar flares and space weather with unprecedented accuracy. Surya's early results beat previous benchmarks by 16%, offering two-hour-ahead predictions that could shield satellites, power grids, and astronauts from solar disruptions.

The model learns directly from high-resolution imagery of the Sun without extensive labeling, adapting easily to multiple heliophysics tasks. Surya’s open-source release on Hugging Face and GitHub invites global researchers to expand its capabilities. It marks a pivotal shift toward AI-powered space weather forecasting—and a broader AI push in planetary and Earth sciences. → Read the full paper here.

🛰️ NEWS

What Else is Happening

🤥 AI Browsers Fooled by PromptFix: A new exploit hides malicious prompts in fake CAPTCHAs, tricking AI agents like Comet into making purchases or entering credentials—no user input needed, just invisible instructions.

🎧 Pixel Buds Get Smarter with Gemini: Google’s new Pixel Buds 2a and Pro 2 bring AI noise filtering, hands-free Gemini access, and even head-gesture call controls—Pro 2 features arrive via update in September.

🔠 Google Translate Tests Language Lessons: A hidden “Practice” feature adds gamified, AI-powered learning to Translate, signaling a major pivot toward competing with Duolingo and reshaping language education.

🔲 NVIDIA Plans Faster AI Chip for China: A new B30A chip based on Blackwell tech may outperform the H20, as NVIDIA eyes exports amid tight U.S. restrictions and growing China trade tension.

📟 Claude Code Joins Anthropic Enterprise Suite: Anthropic now bundles its popular command-line tool with Claude for Enterprise, offering deeper integrations, admin controls, and better scalability for heavy business use.

🏈 NFL Upgrades Surface Tablets with AI: Microsoft’s new Copilot+ tablets give coaches instant access to game data and video, speeding up decisions—just don’t expect AI to call plays (yet).

𝕏 TWEETS

Google Leads AI Hype Poll

📽️ VIDEO

White Collar Jobs, Hyperscalers, AI Coding, Open vs Closed, Agents, and more!

Matt sits down with AWS’s Matt Garman to discuss AI’s impact on white-collar work, 80% of developers using AI, and why adaptability is key to future success.

That’s a Wrap!

📢 Want to advertise with Forward Future? Reach 550K+ 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

🧑‍🚀 🧑‍🚀 🧑‍🚀 🧑‍🚀