Good morning. It's Wednesday, January 28, and we're covering AI worship fears, OpenAI’s talent shakeup, UK’s Anthropic pilot, and more.

First time reading? Join a community of 600k+ people keeping up with the future of tech: Sign up here. If you’re enjoying this newsletter, share it with a friend. And as always, let us know if you have any feedback: [email protected]

🤖 First Full‑Body Brain for Humanoid Robots
Figure AI introduced Helix 02, an upgraded version of its humanoid robotics AI that extends autonomous control from just the upper body to the robot’s entire body—including walking, balancing, and manipulation—through one unified neural system. Helix 02 enables continuous, long‑duration tasks, such as autonomously unloading and reloading a dishwasher across a full kitchen, without resets or human input.

🤝 UK Taps Anthropic for AI Assistant Pilot
Anthropic has been selected to build a government AI assistant to modernise how citizens navigate complex UK state services. The pilot, led by the Department for Science, Innovation, and Technology (DSIT), focuses on agentic AI that guides users through processes rather than just providing information. Initial deployment targets employment services, helping users find work, training, and support using a "Scan, Pilot, Scale" safety framework.

👨‍🔬 OpenAI Launches Prism for Scientists
OpenAI introduced Prism, a free AI-native workspace for scientific writing and collaboration powered by the new GPT-5.2. Available January 27, 2026, Prism integrates drafting, LaTeX editing, citations, and collaboration into a single cloud-based tool. It builds on the Crixet platform and aims to accelerate scientific progress by handling the "mechanical parts of research." It is currently open to personal ChatGPT accounts, with organizational access coming soon.

🚀 Moonshot Launches Open Kimi K2.5
Moonshot AI released Kimi K2.5, an open-source multimodal model trained on 15 trillion tokens, alongside a new coding agent called Kimi Code. The company reports the model outperforms GPT-5.2 and Gemini 3 Pro on several coding and video benchmarks, featuring a capability dubbed "vibe coding" (video-to-code). The launch follows news that Moonshot is finalizing a fresh funding round at a $4.8 billion valuation.

CompactifAI by Multiverse Computing compresses LLMs by up to 80% while maintaining accuracy, delivering up to 50% cost and energy savings for companies like IBM, Bank of Canada, and Moody’s. Run your models on fewer GPUs or even on the edge when you deploy CompactifAI to scale your AI workloads!

Try CompactifAI for Free and Realize Immediate Savings!

Quick test. One image. Two possibilities. Let’s see if you can still tell what’s real. Is this image real or fake? Answer at the bottom! 👇

OpenAI’s CFO sparked debate at Davos after suggesting value-sharing deals in drug discovery, raising fears about IP, compute power, and who profits from AI breakthroughs when the model contributes to the invention.

In an essay for The New York Times, Joseph Bernstein argues that AI is being draped in religious imagery, evolving from a mere tool into a modern deity complete with its own prophecies and "virgin birth" narratives. This cultural shift, highlighted by figures like Joe Rogan and scholars like Lucy Suchman, speaks to a deep-seated human tendency to anthropomorphize "black box" systems. When we treat chatbots as oracles, we transition from technical skepticism to a faith-based trust in systems whose inner workings are intentionally kept opaque.

Ultimately, this deification serves a very terrestrial purpose: it reinforces the authority of the tech elites who build these systems. By framing AI as a transcendent or messianic force, the corporations behind the curtain become unquestionable. The real concern isn't the "god" itself, but the "priests" at the helm who get to write the digital scripture, control the narrative, and profit from our collective devotion. → Read the full article here. (Paywall)

Thinking Machines Lab, once a $12 billion Silicon Valley darling, recently buckled under a high-stakes internal rift between CEO Mira Murati and her technical core. The conflict centered on a stalled $50 billion valuation and a disagreement over whether to remain independent or seek a strategic partner. The tension reached a breaking point on January 14, when Murati fired CTO Barret Zoph, triggering a mass exodus of key talent—including co-founders Luke Metz and Sam Schoenholz—back to the familiar arms of OpenAI.

This collapse underscores how fragile even the most well-funded AI startups can be when personal loyalties and governance collide. Between Meta’s "Godfather" compensation packages and OpenAI’s aggressive re-absorption of talent, the path to independence is narrowing. As capital tightens, the industry is shifting away from a diverse ecosystem of new institutions and toward a consolidation of power among a few dominant, incumbent players. → Read the full article here. (Paywall)

Alibaba announced Qwen3-Max-Thinking, describing it as its most capable reasoning model to date. The company says the model was trained at massive scale and is designed to excel across reasoning, knowledge, tool use, and agent-style tasks.

Alibaba highlighted a new mechanism called "Experience Extraction" and "Test-time Scaling," where the model identifies and removes redundant logical paths in real-time to improve efficiency. Reported benchmarks show it achieving a score of 58.3 on Humanity's Last Exam (HLE), significantly outperforming GPT-5.2-Thinking (45.5) and positioning it as a global leader in reasoning.→ Read the full article here.

©️ Clawdbot Renamed After Trademark Push: Creator Peter Steinberger confirmed Anthropic forced his viral, open-source AI agent to rebrand as Moltbot due to trademark issues.

👤 Airtable Launches Superagent AI Platform: CEO Howie Liu positions Superagent as a multi-agent system that coordinates specialized AIs in parallel to deliver interactive, data-rich outputs.

🤔 Google Health Concerns: A German study of 50,000 searches found Google’s AI Overviews cite YouTube (4.4%) more than any single medical authority, raising reliability concerns.

🐾 AI Tracks Dino Footprints: Scientists launched DinoTracker, an AI tool that identifies dinosaurs from footprints with 90% accuracy, linking tracks from 200 million years ago to modern bird evolution.

The Image Was Real. Tune in next week for another Real or AI test.

❤️ Love Forward Future? Share your referral link with friends and colleagues to unlock exclusive perks! 👉 Get your link here.
🧑‍💻The AI-Fluent Job Board: Hire talent that understands AI. Find jobs that actually value your AI skills. → Learn more.

Thanks for reading today’s newsletter. See you next time!
Matthew Berman & The Forward Future Team

🧑‍🚀 🧑‍🚀 🧑‍🚀 🧑‍🚀 🧑‍🚀