Good morning, it’s Tuesday. Waymo may soon use interior cameras to train generative AI (and serve you ads), AI agents are learning to hack, and China’s building self-critiquing models that might outreason the West’s finest. Oh—and Peter Diamandis says you just have to survive five more years to live forever.
Read on!
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
Tariffs Test the Global AI Supply Chain
In early April, President Donald Trump imposed sweeping new import tariffs that have sent ripples through the global technology market. The policy includes a 10% baseline tariff on all imports into the U.S., with "reciprocal" duties as high as 32% on goods from specific countries that run large trade surpluses with the U.S. → Continue reading here.
🛡️ CYBERSECURITY
AI Agents Poised to Transform Cybercrime as Autonomous Attacks Loom
The Recap: AI-powered agents are emerging as a serious cybersecurity threat, with researchers warning that they could soon carry out cyberattacks autonomously and at scale. These systems, once developed to automate benign tasks, now pose risks as they become cheaper, faster, and more adaptive than human hackers. In MIT Technology Review, journalist Rhiannon Williams explores early warning efforts and the race to detect these agents before they go mainstream.
Highlights:
AI agents are more capable than traditional bots, able to adapt, plan, and avoid detection during cyberattacks.
Palisade Research’s “LLM Agent Honeypot” project has logged over 11 million access attempts, identifying eight likely AI agents, with two confirmed—originating from Hong Kong and Singapore.
Researchers use prompt-injection tests and timing analysis to distinguish AI agents from humans and simple bots.
Mark Stockley of Malwarebytes warns we may soon live in a world where most cyberattacks are carried out by agents.
AI agents could enable scalable ransomware operations by automating tasks like target selection.
A benchmark created by Daniel Kang at the University of Illinois found that AI agents exploited up to 25% of vulnerabilities with minimal information.
Forward Future Takeaways:
AI agents bring a new level of sophistication to cyber threats, moving beyond brute-force scripts to context-aware, autonomous operations. As defenses evolve, the cybersecurity community faces a dual challenge: tracking these agents before they scale, and using AI itself to preemptively fortify systems. The real test ahead is whether we can build resilience faster than bad actors can build bots. → Read the full article here.
🧑🏫 FORWARD FUTURE PRO
In case you missed it: I recently shared my process for using Midjourney to place an existing character—our Forward Future astronaut, Astro—into new poses and scenarios. You’ll learn how to craft precise prompts that adjust posture, mood, and dynamics while keeping the character’s core style intact. This is especially useful for storytellers, content creators, and anyone who needs consistent character portrayals across scenes. → Premium subscribers can read it here.
🤖 MODELS
China’s DeepSeek Unveils Self-Improving AI Model to Compete with GPT-4o and Gemini
The Recap: Chinese AI startup DeepSeek, in collaboration with Tsinghua University, has developed a method to make language models smarter and more efficient using self-improvement techniques. The new system, called DeepSeek-GRM, uses a built-in AI “judge” and a process called self-principled critique tuning (SPCT) to enhance the model’s reasoning. The team claims it could outperform top competitors like OpenAI’s GPT-4o, Google’s Gemini, and Meta’s LLaMA.
Highlights:
DeepSeek’s new approach improves large language model (LLM) reasoning without scaling up model size.
The core technique, SPCT, allows the AI to develop and apply its own criteria for judging its answers.
DeepSeek-GRM evaluates answers in real-time, providing feedback based on how closely outputs align with defined principles.
This generative reward modeling (GRM) approach is more compute-efficient than traditional scaling.
DeepSeek says its models outperform GPT-4o, Gemini, and LLaMA in general queries, though independent benchmarks have not been shared.
The company plans to release its models as open-source, but no timeline has been announced.
Rumors suggest a successor chatbot, R2, is imminent, but DeepSeek has not confirmed a release.
Forward Future Takeaways:
DeepSeek’s self-judging model architecture signals a shift in AI development—from brute-force scaling to smarter, principle-based refinement. If effective, GRM and SPCT could offer a path to more capable, efficient, and accessible AI systems. The big question: can these methods scale across tasks without unintended side effects or bias reinforcement? → Read the full article here.
🛰️ NEWS
What Else is Happening
👁️ Google Adds Visual Search to AI Mode: Snap a pic, ask a question—AI Mode now gets what you see, thanks to Gemini’s multimodal power.
💰 OpenAI Eyes AI Device Merger: OpenAI explores buying Sam Altman and Jony Ive’s mystery hardware startup to bring it fully in-house.
💸 IBM Buys Hakkoda to Boost AI Consulting: The Snowflake partner adds deep data chops as IBM preps clients for AI-driven transformations.
🏞️ Google AI Mode Gets Visual Upgrade: Users can now ask smart questions about images—AI Mode breaks down scenes using Lens and Gemini tech.
💺 Porsche Uses AI to Judge Ride Comfort: A neural net now mimics human testers to evaluate chassis feel—speeding tuning and possibly riding in future cars.
✨ Galaxy S25 Gets Gemini Live AI: Samsung’s latest update lets users chat with AI in real time using their camera—style tips and space hacks now visual.
📽️ VIDEO
LLaMA 4 is HERE! Meta Just COOKED
Meta drops Llama 4: multimodal, open-weight, and boasting a wild 10M-token context window. Three sizes, record-breaking benchmarks, and a Behemoth model still cooking. Get the full scoop in Matt’s latest video! 👇
🧰 TOOLBOX
Smarter Payments, Effortless Video Creation, and AI-Powered Job Hunting
💳 Chargeblast: Prevent chargebacks with real-time alerts, issuer data, and automated tools—cut disputes by up to 90%.
🎥 SkyReels AI: Effortlessly create high-quality videos with AI-driven tools for scriptwriting, storyboarding, dubbing, and editing—all in one platform.
💼 Careerflow AI: Streamline your job search with AI-powered tools for resume building, cover letter generation, LinkedIn optimization, and application tracking.
🗒️ FEEDBACK
Help Us Get Better
What did you think of today's newsletter? |
Reply to this email if you have specific feedback to share. We’d love to hear from you.
🤠 THE DAILY BYTE
Fake News, Real Consequences: AI Creator Behind Trump’s Gaza Video Says ‘Ban the Bots’ in Politics
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.
📥 Got a hot tip or burning question? Drop us a note! The best reader insights, questions, and scoops may be featured in future editions. Submit here.
🛰️ 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
🧑🚀 🧑🚀 🧑🚀 🧑🚀
