AI Newsletter December 2025

Welcome to the Dixon AI Newsletter which summarises selected AI stories and releases you might have missed. This issue covers AI news from 22nd November to 15th December 2025.

The AI Macro Environment and Enterprise Insights

• Reports galore! As we approach the end of the year, many companies have been putting out reports detailing the current state of AI usage. 

  • Open AI published The State of Enterprise AI report (Open AI).  This report details how enterprise AI adoption is rapidly scaling and deepening into core workflows, highlighting significant productivity gains and a widening gap between industry leaders and laggards.

  • Microsoft published The Copilot Usage Report 2025 (Microsoft). Analyzing 37.5 million conversations, this report reveals distinct behavioral patterns where desktop usage mirrors professional work rhythms while mobile usage functions as a personal companion for health and introspection.

  • Menlo published a 2025: The State of Generative AI in the Enterprise (Menlovc.com). This market analysis tracks the surge in enterprise AI spending to $37 billion, contrasting the dominance of startups in application layers like coding against incumbents' strength in infrastructure.

  • MIT Sloan and BCG published The Emerging Agentic Enterprise: How Leaders Must Navigate a New Age of AI (MIT / BCG). This research explores the management challenges of "Agentic AI," urging leaders to redesign governance and workforce roles to navigate the strategic tensions of using systems that function as both tools and autonomous coworkers.
    

• Agentic AI Foundation: The Agentic AI Foundation (AAIF) is a newly launched open-source initiative hosted by the Linux Foundation and co-founded by industry leaders like OpenAI, Anthropic, and Block designed to create a shared "language" that allows different AI systems to work together safely and effectively. Its purpose is to establish neutral standards and protocols ensuring that powerful, autonomous AI agents evolve as an open, interoperable ecosystem rather than fragmented, incompatible tools. Agentic AI Foundation (AAIF) 
  

• Trump Signs Executive Order to Pre-empt State AI Laws: Moving from threat to action, President Trump signed an executive order on December 11 officially establishing a federal AI Litigation Task Force. The new body, housed within the DOJ, is empowered to sue states like California and Colorado over safety regulations the White House deems "onerous" to industry growth. The order also threatens to withhold federal broadband funding from non-compliant states, setting the stage for an immediate constitutional showdown between Washington’s deregulation agenda and state-level consumer protections. See https://www.whitehouse.gov/presidential-actions/2025/12/eliminating-state-law-obstruction-of-national-artificial-intelligence-policy/.
  

• Anthropic IPO? Anthropic are reportedly preparing for a massive IPO in 2026 which could test AI valuations and fears of a bubble.  Anthropic reportedly preparing for one of the largest IPOs ever in race with OpenAI: FT 
  

• OpenAI and Disney: Open AI and Walt Disney reach an agreement to bring Disney’s brands to Sora. The Walt Disney Company and OpenAI reach landmark agreement to bring beloved characters from across Disney’s brands to Sora 
  

New and updated AI model and application releases

• OpenAI 

  • GPT 5.2: OpenAI released GPT-5.2 to all ChatGPT users. GPT 5.2 reportedly boasts a 30% reduction in hallucinations, and significant improvements in long-context understanding and multi-step reasoning. Introducing GPT-5.2 | OpenAI. Also see the official GPT-5.2 Prompting Guide.

  • Expanded Data Residency: Open Ai have expended their data residency for business customers, giving organisations more control over their data Expanding data residency access to business customers worldwide | OpenAI 
    

• Google

  • Deep Research Agent: Google released a significant update to their deep research agent, producing higher-quality output with reduced hallucinations at a lower cost. Build with Gemini Deep Research 

  • Deep Think: Gemini Deep Think, the world's most powerful thinking model, is out now to Google AI Ultra subscribers. Gemini 3 Deep Think is now available 

  • Stitch: Google has released a new experimental AI-powered design tool. Bring your app ideas to life with Gemini 3 in Stitch. 
    

• Microsoft

  • Agent Dashboard: Microsoft is rolling out a new dashboard for tracking agent usage across organisations. New! Centralized Agent Dashboard and Enhanced Reporting | Microsoft Community Hub 
    

• Anthropic

  • Opus 4.5: Anthropic released their most powerful model to date. Opus 4.5 is particularly strong at coding, deep research and working with slides and spreadsheets. Introducing Claude Opus 4.5 \ Anthropic

  • Claude for Excel: The Excel plugin beta has been rolled out to all Max, Teams and Enterprise users Claude for Excel 
    

• Notable Other Model Releases

  • Mistral 3: Mistral, owners of the leading European model, released Mistral 3, the leading open family of models. Introducing Mistral 3 
    

• Notable Application Releases

  • Runway Gen-4.5: The new video generation model from Runway, Gen-4.5, is the world’s leading video generation model. https://runwayml.com/research/introducing-runway-gen-4.5 

  • Marble: A new  World Model enables users to create 3D worlds from a single prompt: Marble 
    

Recommended Reading, Listening and Book Releases

• For anybody:

  • Video: Watch ‘The Thinking Game,’ a documentary about Google DeepMind, for free on YouTube. 

  • Podcast: AI in the UK. Check out the new podcast series from our very own Niels Footman, focusing on everything happening with AI in the UK.  AI in the UK 
    

• Coming soon!

  • Book: The Dixon AI Transformation Playbook will be published later this month (December 2025) in both paperback and ebook formats. 
    

From Dixon AI

• AI Drop In: one-hour, live session held three times a week. Join here: https://www.dixonai.com/ai-drop-in 

  • w/c 15th Dec: What’s New in AI

  • The Drop-in sessions will then be taking a break over the end of year period and will start again on Monday 12th January.
    

• New Website: We have a new website launching imminently - keep an eye on www.dixonai.com over the coming days.
  

A word from Rob

2026 is the last year to be early with AI

As 2025 draws to a close, we have just passed an important milestone. It is now three years since the launch of ChatGPT. In that short time, AI has moved from curiosity to capability, and from experimentation to real commercial impact.

My strong observation is that we are now beginning to enter the steeper part of the S-curve. Progress is no longer incremental. Capability is accelerating, adoption is spreading, and competitive advantage is beginning to compound. This is the phase where change becomes unavoidable.

For tech-first sectors, the implications are already obvious. If you are not actively adopting and optimising AI, you are already late. The baseline for competition in those markets has already shifted.

However, for many other sectors, manufacturing, professional services, recruitment, education, public services, healthcare, legal, accountancy, scientific research and managed services, there is still time to act. But that window is narrowing quickly.

In my view, 2026 is the last year to be early with AI.

By 2027, many organisations will be facing competitors in their own industries who have successfully embedded AI across their value chains. These organisations will be delivering enhanced products, enhanced services and enhanced levels of value, at a speed and cost base that materially changes the competitive landscape.

At that point, those who delayed will be forced to operate defensively, reacting to disruption rather than shaping it.

So what does this mean for owners and senior executives right now?

First, this is not a technology problem to be delegated. AI is a core commercial issue. It should not be outsourced, and it should not sit solely with the IT function. Leaders need direct exposure to the tools and a clear understanding of what is now possible, because strategic decisions about markets, offerings and operating models will increasingly depend on that understanding.

Second, focus on the who before the what. Your priority in 2026 should be building internal capability. Upskill yourself. Identify the people in your organisation who have the mindset to lead change. Back them, equip them properly and give them permission to innovate.

Finally, prioritise momentum over perfection. Real progress comes from doing. Practical projects, grounded in commercial reality, build learning and confidence far faster than isolated pilots or lengthy strategy documents.

2026 is the last year to be early. After that, the same actions will still be required, but they will be taken under far greater pressure and with far fewer options.

This is a genuine opportunity for those willing to act decisively, and a clear warning for those who choose to wait.