Welcome to the latest edition of The Neural Link!
In this edition, we track the biggest leaps and shake-ups from the past month. Generative AI is entering the year in a very different place. Trillion-dollar mergers, courtroom battles over training data, and shifting power inside Big Tech point to an industry moving from experimentation to consequence. Innovation is still accelerating, but questions of control, legality, and scale are now shaping the direction of travel.
Let’s dive in.
This month was quieter on flagship releases, with no major new foundation models from the big players. Instead, January saw targeted model announcements in specialised areas, alongside continued consolidation and product-led evolution across existing frontier models.
|
Company |
Model / Launch |
What’s new |
|
Moonshot AI |
Kimi K2.5 |
A new large-scale Mixture-of-Experts model was announced in January 2026, positioned as a high-performance conversational and reasoning model, with a reported trillion-parameter architecture. |
|
Lightricks |
LTX-2 (open source) |
Lightricks open-sourced its LTX-2 text-to-video generation model, releasing full model weights and making advanced video generation more accessible to developers and creators. |
SpaceX acquires xAI in $1.25 trillion
mega deal
Elon Musk has merged his AI startup xAI into SpaceX, creating the largest technology deal on record. The move brings rockets, data centres, and generative AI under one organisation, with Grok now tied directly to SpaceX’s infrastructure and capital base.
Why it matters:
This deal concentrates enormous technical and financial power in a single private entity. It signals a new phase of vertical integration in AI, where scale, compute, and capital may matter more than model quality alone.
Music publishers sue Anthropic for $3 billion
over AI training data
Major music publishers have filed a lawsuit accusing Anthropic of illegally using more than 20,000 copyrighted works to train its Claude models. The case follows earlier author lawsuits and seeks damages exceeding $3 billion.
Why it matters:
Courts are beginning to draw sharper lines between lawful AI training and unlawful data acquisition. The outcome could reshape how AI models are trained and how enterprises assess legal risk when adopting generative tools.
Apple acquires Israeli audio AI
startup Q.ai
Apple confirmed its acquisition of Q.ai, a startup specialising in “silent speech” technology that interprets facial micro movements to detect whispered or unspoken words. The deal is Apple’s second-largest acquisition to date.
Why it matters:
Apple appears to be investing beyond chatbots and voice commands toward ambient, hands-free AI. This points to a future where AI interaction becomes passive, contextual, and always present rather than explicitly prompted.
Nvidia CEO denies tension with OpenAI
Nvidia CEO Jensen Huang dismissed rumours of strain with OpenAI following reports that a $100 billion funding deal had stalled. The comments come as Nvidia’s influence over the AI ecosystem continues to grow through its control of critical compute.
Why it matters:
The AI race is increasingly shaped by who controls infrastructure rather than just who builds models. Shifts in these relationships can ripple quickly across pricing, access, and competitive dynamics.
Google tests AI-generated 3D worlds with Project Genie
Google DeepMind’s Project Genie allows users to generate short, playable 3D environments from text prompts. While still limited and inconsistent, early demos show AI moving beyond static content into interactive experiences.
Why it matters:
Generative AI is expanding into simulations, games, and training environments. These capabilities could have long-term implications for education, design, and scenario modelling well beyond entertainment.
Other news in AI
-
OpenClaw’s AI assistants are now building their own social network
Open source AI agents are self-organising on a Reddit-like platform called Moltbook, showing how autonomous systems can interact and coordinate without direct human oversight. -
There’s a social network for AI agents, and it’s getting weird
More than 30,000 AI agents are posting and forming communities autonomously, raising questions about emergent behaviour and responsibility in agent-driven systems. -
“Stealing Isn’t Innovation” Hundreds of creatives warn against an AI-slop future
Artists, writers, and musicians are calling for stricter rules on AI training data, warning that unchecked scraping threatens creative industries and cultural quality. -
Firefox will soon let you block all of its generative AI features
Mozilla announced a single toggle to disable AI features in Firefox, responding to growing user demand for transparency and control. -
VCs say consumer AI startups still lack staying power
Investors argue that most consumer AI apps struggle to retain users, while enterprise-focused AI companies continue to attract long-term capital. -
Google DeepMind hires voice AI startup team in acqui hire
DeepMind quietly hired key staff from Hume AI to strengthen Gemini’s voice capabilities, highlighting the growing use of talent deals over full acquisitions. -
Meta halts teen access to AI chat characters
Meta paused minors’ access to its AI personas across Instagram, Facebook, and WhatsApp following safety concerns around generative chat experiences.
This is the first signal of what’s shaping up to be a defining year for generative AI. Power is concentrating, guardrails are being tested, and the gap between hype and reality is narrowing fast.
We’ll be tracking what matters as the year unfolds.
If you’re thinking about what this means for your organisation, let’s talk. And if you haven’t already, subscribe to The Neural Link for a monthly view of the trends shaping AI and automation, delivered straight to your inbox.
