Today's Highlights:

• EU vs Meta: WhatsApp’s AI Walled Garden
  The European Commission has launched a formal antitrust investigation into Meta’s decision to block third-party AI chatbots from WhatsApp’s API while keeping only its own Meta AI assistant live. Regulators worry this could shut out competitors like other LLM providers and entrench Meta’s control over a critical messaging rail used by billions. Financial Times · The Verge Financial Times+1

• Amazon’s Nova 2 Omni: One Model to Rule Your Data
  Amazon introduced Nova 2 Omni, a multimodal model on Bedrock that can process long-form text, images, video, and speech together—up to hundreds of pages or hours of media—and generate both text and images from a single context window. For enterprises, the pitch is simpler AI stacks (one model instead of many) powering agents that can understand product catalogs, documents, and video libraries in one go. Amazon About Amazon

• DeepSeek V3.2: Frontier Models on a Budget
  DeepSeek’s new V3.2 and V3.2-Speciale models posted eye-catching scores on math contests like AIME 2025 and the Harvard-MIT Mathematics Tournament, edging out GPT-5 High and Gemini 3.0 Pro in some benchmarks while reportedly running at a fraction of their cost. That combo of high-end performance and price pressure is exactly what enterprises and governments have been waiting for in the post-frontier-model era. VentureBeat Venturebeat

• Cursor Joins the “Decacorn DevTool” Club
  AI coding startup Cursor, maker of an AI-native IDE, closed a massive $2.3B Series D round at a $29.3B valuation, with backing from Accel, Thrive, a16z, DST, Nvidia and Google. The raise signals that investors still see huge upside in “AI as co-developer” tools—especially those embedded directly in engineers’ daily workflows. Intellizence · Crunchbase News Intellizence |+1

• Vertical AI in Law and Cyber: Harvey & 7AI Break Out
  Legal AI platform Harvey raised a $160M late-stage round at an $8B valuation, funding expansion of its case-law-aware models and collaborative “Shared Spaces” for law firms, while cybersecurity startup 7AI landed $130M Series A to deploy agentic AI that triages and responds to security alerts in real time. The takeaway: specialized AI agents for high-stakes domains are firmly in hypergrowth mode. SiliconANGLE · Wall Street Journal SiliconANGLE+1

Deep Dive:

Deep Dive #1 – Nova 2 Omni & the Agentic Enterprise

Amazon’s Nova 2 Omni marks a clear step toward “do-it-all” enterprise models: instead of juggling separate text, image, and video systems, Nova promises unified multimodal reasoning inside Bedrock. It can ingest entire product catalogs, knowledge bases, and media archives, then power agents that answer questions, generate content, and orchestrate workflows across that data. Amazon About Amazon

For founders and product teams, the implication is that “agentic AI” may increasingly be built on top of fewer, more capable foundation models—where differentiation shifts to data, orchestration, and UX. If Nova performs as advertised at scale, you’ll see more companies standardize on a single provider for multimodal workloads, while using lighter models only at the edges for latency or cost reasons.

On the infra side, Nova underscores a broader shift: hyperscalers want to own not just compute, but the model layer and the agent frameworks that run on top of it. That’s good news for teams already inside AWS’s ecosystem, but it raises strategic questions for those trying to stay cloud-agnostic or mix-and-match best-of-breed models. Expect a lot more experimentation with multi-cloud and “bring-your-own-model” setups as enterprises weigh lock-in vs. speed.

If you’re building AI tools, the near-term takeaway is tactical: design your products assuming customers will eventually want end-to-end, multimodal flows—text + docs + calls + screen recordings—rather than siloed text chat. Nova’s feature set is a clear signal of where enterprise expectations are heading. AWS re:Invent coverage SiliconANGLE

Deep Dive #2 – The WhatsApp AI Antitrust Battle

The EU’s new antitrust probe into Meta’s WhatsApp AI policies is about more than one chatbot—it’s a test case for how regulators will treat AI access on dominant platforms. Meta recently updated WhatsApp’s terms to block new third-party AI providers from using its API (and will phase out existing ones), effectively giving exclusive real estate to its own Meta AI assistant. Financial Times · Reuters Financial Times+1

For European regulators, the concern is familiar: if one company controls both the messaging rail and the default AI assistant on that rail, it can starve rivals of distribution, data, and user feedback loops. This is especially sensitive in AI, where scale advantages compound quickly. The investigation will look at whether Meta’s policies unfairly restrict other AI tools from reaching users, potentially violating EU competition law even beyond the new Digital Markets Act. The Verge The Verge

If the EU pushes back hard—forcing interoperability, mandating fair access, or even imposing structural remedies—it could set a precedent that other jurisdictions emulate. Messaging platforms, app stores, and productivity suites might then be expected to offer “open slots” for AI providers, similar to browser and search choice screens in the past.

For builders, this case is worth watching closely. It may determine whether the primary distribution channel for consumer AI assistants is a handful of vertically integrated platforms, or a more open ecosystem where users can easily plug in competing agents into WhatsApp, SMS, email, and beyond.

Global AI News

• EU’s Digital Omnibus & GDPR Tune-Up for AI
  European regulators rolled out a “Digital Omnibus” package that proposes streamlining GDPR compliance and clarifying rules around automated decision-making and AI-driven services, alongside fresh guidance on how the AI Act applies to General-Purpose AI models. For AI builders in Europe, this means more documentation, clearer risk-classification rules, and likely new obligations for GPAI transparency. MoFo · White & Case Morrison Foerster+1

• Data Protection Round-Up: India’s First Digital Privacy Law Lands
  A November data-protection briefing highlighted India’s finalization of its first comprehensive digital privacy law, which will shape how global AI companies collect and process user data in one of the world’s largest internet markets. The update also flags new EU data-access rules under the DUAA, tightening governance around data sharing for AI. Stephenson Harwood shlegal.com

• Africa’s AI Moment: Green Compute as a $1.5T Bet
  The World Economic Forum spotlighted how coordinated investment in low-carbon data centers and “green compute” could unlock up to $1.5T in value for Africa, arguing that energy-efficient AI infrastructure is the key to leapfrogging legacy tech and enabling homegrown AI ecosystems. World Economic Forum World Economic Forum

• LatAm & Africa: AI-Powered Commerce, Cities, and Funding
  Chilean AI startup Vambe raised $14M to automate conversational commerce for mid-sized B2C brands across Latin America, while the African Cities Innovation Fund launched grants of up to $75K for digital and AI-enabled solutions for urban transport, climate resilience, and basic services—signaling growing support for applied AI in emerging markets. SiliconANGLE · TechCabal SiliconANGLE+1

Market Trends:

• AI Megarounds Are Back (If They Ever Left)
  November funding data shows AI still dominating late-stage capital: Cursor’s $2.3B Series D, major rounds for AI data-center provider Lambda ($1.5B) and prediction-markets platform Kalshi ($1B) anchored another month of mega-deals, pushing AI to the center of global VC activity. Intellizence · Crunchbase News Intellizence |+1

• Vertical AI: Legal Tech’s Funding Frenzy
  Harvey’s $160M round at an $8B valuation caps a breakout year for legal-AI startups, with 2025 seeing billions flow into tools that draft, review, and analyze contracts and case law. A recent survey of legal-tech deals notes that investors increasingly prefer domain-expert AI over generalist copilots, betting that workflows in law, tax, and compliance will be among the earliest to fully industrialize AI. SiliconANGLE · Business Insider SiliconANGLE+1

• Security Agents: 7AI’s $130M Signal
  7AI’s $130M Series A—unusually large for such an early stage—highlights investor conviction that SOCs will need autonomous AI agents to handle the flood of security alerts. As attackers adopt AI for more sophisticated campaigns, capital is quickly chasing startups that can combine LLMs with traditional security telemetry to cut response times from hours to seconds. Wall Street Journal The Wall Street Journal

• Fintech & Wealth: AI Eats the Back Office
  In financial services, AI-native players are picking up momentum: Flex raised a $60M Series B to give “jumbo shrimp” mid-market businesses an integrated AI-powered finance stack; Range secured $60M Series C to scale AI-driven wealth management; and Nevis emerged from stealth with $40M to build an AI platform for advisors. Together they show how AI is moving into core revenue and compliance workflows in finance, not just chatbots and ops. Reuters · Cathay Capital · FF News Reuters+2Cathay Capital+2

• Early-Stage Buzz: PR, Enterprise Ops, and Agentic Automation
  On the earlier-stage side, Honeyjar AI raised $2M pre-seed to build an AI workspace for PR and communications teams; CoreOps.AI closed $3.5M to modernize legacy enterprise systems with AI-driven ops; and new seed-stage deals in “agentic automation” for regulated industries continue to pop up weekly—evidence that founders see huge whitespace in deeply embedded, workflow-specific AI agents. Axios · Economic Times · SiliconANGLE Axios+2The Economic Times+2

Want to advertise?