In December 2025, Sam Altman at OpenAI called a “code red” as Google’s tools started catching up. And this is for good reason. They have a number of great tools. Quietly, methodically, Google’s AI stack had caught up. And in some (many?) places, it may now be ahead.

For media leaders, this matters. Not because Google suddenly learned how to do AI — it has been building foundational models, infrastructure, and distribution for more than a decade — but because those pieces are lining up into tools that are usable, fast, and increasingly invisible. The challenge, as ever with Google, is working out what’s what. Branding clarity has not, in my opinion, been its strength. 

Gemini: which is at the center of everything, the brain behind Google’s AI ecosystem rather than a single product. Gemini 2.0 has real-time reasoning, agentic workflows, and multimodal control, allowing systems to read, watch, listen, and act in concert.

Gemini Nano: shifts the game again by running directly on-device. On supported Android and Pixel devices, this means transcription, summarisation, translation, and analysis without sending sensitive material to the cloud. For newsrooms handling embargoed investigations, source protection, or regional reporting in low-connectivity environments this could be extremely useful. 

Veo: Google’s advanced video generation model is emerging as a credible competitor to OpenAI’s Sora. Together, they point toward faster visual prototyping, explainer creation, and social-first storytelling without the latency and cost that previously slowed experimentation.

“Anti-Gravity” initiative (internal name): the ambition to make AI feel instant, weightless, and invisible. This isn’t a product so much as an infrastructural philosophy which brings many things together, including Android distribution and real-time multimodal systems. The result is AI that disappears into the workflow.

We’re getting into the practical stage of AI. Real-time translation and live captioning become defaults, not features. Automated clipping and highlight generation can happen at broadcast speed. Personalisation — long promised and rarely delivered well — becomes easier to deploy at scale because latency and cost barriers shrink. This is how AI-native broadcast and streaming pipelines stop being future road maps and start being operational realities.

Of course, none of this exists in isolation from Google’s search experience. AI Overviews (AIOs) continue to evolve daily, reshaping discovery, referral traffic, and audience expectations. While many of these capabilities originated in DeepMind, they are now being consolidated under the Gemini umbrella — a signal Google is serious about presenting a unified AI platform, even if the naming still lags behind the ambition.

So what else should media companies be looking at beyond Gemini itself?

NotebookLM is one of Google’s most underrated tools for journalists and editors in my opinion. It’s a research assistant that grounds answers strictly in your own documents. For investigations, policy reporting, or long-form projects, it offers a controlled alternative to open-ended chat tools. I’ll cover this in more detail soon.

Vertex AI is where publishers with in-house data teams should focus. It provides production-grade model hosting, fine-tuning, and evaluation — essential for organisations building proprietary recommendation systems, internal copilots, or audience intelligence tools.

Finally, Google’s agent frameworks — increasingly embedded across Workspace and Gemini — are worth watching closely. The shift from AI that answers questions to AI that executes tasks is where real operational leverage emerges.

Google hasn’t suddenly become interesting. It has become coherent. And for media companies navigating the transition from search to answers to agents, this matters more now than ever before.

Google and the media have a long, complicated history. Search is decreasing and we have to rethink distribution in an AI world. Google is still one of the, if not the, largest funder of journalism. Now their AI tools are catching up to give a competitive edge.

Overreliance on a single company is never a good thing, as we have found to our peril, so this competition of AI tools — not just between Google and OpenAI — should be a good thing for the foundation of journalism.

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .