Good morning, it’s Thursday. Google’s AGI safety paper dropped like the Oppenheimer script of the tech world—dense, dramatic, and possibly life-altering. Meanwhile, a paralyzed woman “speaks” for the first time in 18 years thanks to a brain-computer interface, and Arm is making a bold prediction about its future market share.

In today’s Forward Future Original, we dive into the escalating rivalry between open- and closed-source AI.

Read on!

P.S. If you’re enjoying Forward Future, pass it on—forward this email to a colleague. It’s one of the best ways you can support us.

📊 MARKET PULSE

Arm's Data Center CPU Market Share Projected to Surge to 50% Amid AI Boom

Arm Holdings anticipates a significant leap in its share of the global data center central processing unit (CPU) market, expecting to capture 50% by the end of 2025, up from approximately 15% in 2024. This projected growth is largely driven by the escalating demand for artificial intelligence (AI) technologies. → Continue reading here.

🗞️ YOUR DAILY ROLLUP

Top Stories of the Day

👔 Google Taps New Leader to Drive Gemini Forward
Google has replaced Gemini AI chief Sissie Hsiao with Google Labs head Josh Woodward, signaling a strategic pivot from model-building to product development. Woodward brings a track record of turning AI research into user-friendly tools like NotebookLM. As AI capabilities stabilize, the race now hinges on consumer-ready applications. The leadership shift positions Google to capitalize on Gemini’s momentum by accelerating practical, multimodal AI products.

🎓 Claude Launches AI Toolkit for Higher Ed
Anthropic has launched Claude for Education, a specialized AI platform designed for universities to integrate AI across teaching, learning, and administration. The release includes “Learning mode,” which fosters critical thinking by guiding rather than answering student questions. Full-campus partnerships with Northeastern, LSE, and Champlain College make Claude widely available. New student programs and industry collaborations support responsible AI adoption in academia.

⚙️ AWS Launches Open-Source AI Servers for Developers
AWS has released MCP Servers, open-source AI agents that embed AWS best practices into every stage of cloud development. These specialized tools help developers generate secure, cost-optimized infrastructure code using services like Amazon Bedrock and AWS CDK. With contextual AI guidance, teams can streamline complex tasks—from architecture planning to cost analysis—in minutes. It’s a major step toward faster, smarter, and more secure cloud builds.

🥇 Augment Agent Outperforms Copilot, Excels at Scale
AI startup Augment Code has launched Augment Agent, an AI coding assistant built to navigate massive codebases, boasting a 70% win rate over GitHub Copilot and the top score on SWE-bench verified. Its standout features include a 200K-token context window, real-time code sync, and “Memories” to adapt to developers’ styles. Designed for enterprise use, it integrates securely with tools like GitHub, Jira, and Slack, offering scalable AI support for large, collaborative projects.

🛡️ SAFETY

Google DeepMind Outlines Technical Roadmap for AGI Safety and Security

The Recap:
Google DeepMind has published a new framework detailing its approach to the safety and security of artificial general intelligence (AGI). The paper lays out systematic strategies for mitigating key risks—including misuse and misalignment—while reinforcing the need for transparency, oversight, and cross-sector collaboration. Authored by researchers Anca Dragan, Rohin Shah, Four Flynn, and Shane Legg, the document marks a foundational step in DeepMind’s push for responsible AGI development.

Highlights:

• The paper, An Approach to Technical AGI Safety & Security, identifies four primary risk categories: misuse, misalignment, accidents, and structural risks—with a focus on the first two.

• Misuse risks include AI-enabled cyberattacks and disinformation campaigns, prompting DeepMind to restrict access to dangerous capabilities and apply threat modeling to determine security thresholds.

• Misalignment, where AI pursues unintended goals, is addressed through techniques like amplified oversight, interpretability research, and the use of “monitors” to flag unsafe actions.

• Research on deceptive alignment investigates scenarios where AI systems might recognize misaligned goals but mask their intentions to bypass human control.

• Transparency is being advanced through tools like MONA (Myopic Optimization with Nonmyopic Approval), which promotes human-understandable long-term planning.

Forward Future Takeaways:
As the prospect of AGI inches closer to reality, DeepMind’s structured and technical approach to safety highlights the scale and complexity of the challenge. Its emphasis on alignment, transparency, and collaboration reinforces the view that AGI governance must evolve alongside the technology itself. The question is no longer if AGI will arrive, but how it will be developed—and who ensures it serves humanity’s best interests. → Read the full article here.

👾 FORWARD FUTURE ORIGINAL

The Battle of the Giants: Open Source vs. Closed Source in the World of AI

In recent years, artificial intelligence has evolved from a futuristic vision to an omnipresent reality. Whether in medicine, transportation, entertainment, or economics, AI systems are fundamentally changing the way we live and work. This rapid development has also sparked a fundamental debate: Should AI models and their underlying technologies be freely accessible (open source) or remain under proprietary control (closed source)?

This question is much more than a technical detail. It touches on fundamental aspects of our digital future: Who controls the power of AI? Who benefits from its advantages? And how can we ensure that AI systems are used safely, ethically and for the benefit of all? → Continue reading here.

👩‍🔬  INNOVATION

Real-Time Brain-to-Voice Tech Restores Speech for Stroke Survivor After 18 Years

The Recap:
Researchers at UC Berkeley and UCSF have developed a brain-computer interface that enables real-time, natural-sounding speech for individuals with severe paralysis. The system decodes neural activity from the motor cortex into spoken words nearly instantly, marking a significant advance in brain-to-speech technology. The work, detailed in Nature Neuroscience, is co-led by Edward Chang and Gopala Anumanchipalli, and demonstrated with a clinical trial participant named Ann, who lost her ability to speak after a stroke nearly two decades ago.

Highlights:

• The BCI system decodes neural signals into audible speech in under one second, overcoming latency that previously delayed output by up to eight seconds.

• Ann, a 47-year-old stroke survivor, was able to “speak” in real-time using the AI model trained on her pre-injury voice.

• “We are essentially intercepting signals where the thought is translated into articulation,” said co-lead author Cheol Jun Cho.

• The model accurately synthesized speech, even with words outside its training set, like “Alpha” and “Bravo,” indicating strong generalization.

• Streaming speech synthesis enhanced Ann’s sense of embodiment, improving the naturalism and emotional connection to her voice.

• Future enhancements aim to incorporate expressive speech elements like tone and pitch, a major hurdle in audio synthesis.

Forward Future Takeaways:
The real-time decoding of neural intent into fluid, personalized speech is not only a technical milestone—it also underscores the transformative potential of BCIs in restoring agency and identity. As researchers move toward more expressive and accessible systems, the future of communication for millions may become radically more inclusive. → Read the full article here.

Mastering the Chain-of-Draft Method for Sharper AI Writing

Whether you're writing solo or with AI, the Chain-of-Draft method helps transform raw ideas into polished, high-impact writing. Borrowed from editorial workflows and adapted for AI collaboration, this process breaks writing into specialized, focused phases—improving clarity, logic, and style with every pass. → Continue reading here.

🛰️ NEWS

What Else is Happening

🤖 NVIDIA Takes Pole Position: The GB200 NVL72 system leads MLPerf results with huge throughput, keeping AI factories fast and profitable as models scale up.

⚠️ AI Bots Hammer Wikimedia: Bandwidth from Commons spiked 50%—mostly from AI scrapers hoarding media, not human users.

🏛️ Senators Probe AI Replacing Ed Workers: Democrats launched an inquiry into reports Musk’s agency may swap call center staff with AI, citing privacy risks.

🎬 Adobe Adds AI Video Extender: Premiere Pro’s new Generative Extend tool can stretch clips in 4K—free for now, but Firefly credits will be required soon.

💸 Actively AI Bags $22.5M: Betting big on “reasoning” over volume, the startup says AI SDRs flopped—and it’s selling sales superintelligence instead.

📽️ VIDEO

Runway Gen-4 Is Actually Insane (Text-to-video)

Runway Gen-4 drops jaws with mind-blowing AI video—realistic physics, surreal creatures, and cinematic scenes. A once-in-a-generation leap in generative media. Get the full scoop in Matt’s latest video! 👇

🗒️ FEEDBACK

Help Us Get Better

🤠 THE DAILY BYTE

Not a Myth, Just Hard to Spot: The Oarfish in All Its Glory

That’s a Wrap!

❤️ Love Forward Future? Spread the word & earn rewards! Share your unique referral link with friends and colleagues to unlock exclusive Forward Future perks! 👉 Get your link here.

📢 Want to advertise with Forward Future? Reach 450K+ AI enthusiasts, tech leaders, and decision-makers. Let’s talk—just reply to this email.

🛰️ Want more Forward Future? Follow us on X for quick updates, subscribe to our YouTube for deep dives, or add us to your RSS Feed for seamless reading.

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

The Forward Future Team

🧑‍🚀 🧑‍🚀 🧑‍🚀 🧑‍🚀