Google DeepMind to take on AI edits in Photoshop
Adobe and Google have expanded their partnership. It was announced on November 10, 2026 at the Adobe MAX conference in Miami Beach. For artists, the main part of the announcement concerns Photoshop. Adobe and Google DeepMind are beginning joint development of AI editing tools, and Photoshop will be the first to receive them. The companies are exploring how to adapt Google's image model for Photoshop so that edits are performed in near real time.
The second half of the deal works in the opposite direction. Adobe's tools are being embedded into Google's services: Acrobat will appear in Gmail and Google Drive, and Adobe's creative tools will come to AI Mode in Google Search. Google's newest image and video models will be added to Firefly, Photoshop, Premiere and Adobe Express.
What Photoshop will get
The work with DeepMind is still at the research stage. In Adobe's wording, it is about a model that "could" deliver near-real-time edits: an artist changes an image with text prompts and picks from several options the one closest to their intent. The companies do not give release dates, the model's name or a description of the interface. So for now it is impossible to say how this feature will differ from generative fill in practice.
Separately from the partnership, Photoshop has already received a major update at MAX 2026. Its contents are laid out in the community announcement:
- Relight lets you add up to three light sources to a shot and adjust lighting, color, intensity and diffusion right on the canvas. In essence, the light in a frame can be corrected after the shoot.
- Generative Restyle turns a generated edit into an editable layer with its own settings, so the result of the generation can be refined by hand.
- Remove related elements deletes an object along with its shadow, reflection, smoke and other related traces.
- AI Assisted Editor accepts edits via text commands, while the advanced tools remain available in Pro Editor mode.
- Prompt to Edit changes an image from a text prompt and shows several options before applying.
- Quick Compare compares the result with the original in split-screen mode or in single view.
- Generative Upscale increases an image's resolution, sharpness and detail.
Google's models are already available, and the list will expand
Adobe customers can already use some of Google's models. These are Gemini Omni, Veo and Nano Banana. New image and video models will add to this set in Firefly, Photoshop, Premiere and Express. This is how Adobe continues its partner model strategy: in Firefly, third-party developers' models work alongside Adobe's own.
The announcement builds on steps already taken. Adobe released Premiere for Android. Together with YouTube, the company created a dedicated space in mobile Premiere for making YouTube Shorts. Photoshop, Lightroom and Premiere have been ported to Googlebooks, and Adobe CX Enterprise Coworker is embedded in Gemini Enterprise.
"The expanded partnership brings together Google's advanced AI with Adobe's leading creative and office tools on a global scale and allows us to offer distinctive capabilities," said Anil Chakravarthy, president of Adobe and the company's incoming CEO.
PDFs from email open in Acrobat
For studios working in Google Workspace, the change affects everyday tasks. A brief, a contract or a spec in PDF arriving in Gmail usually has to be downloaded and opened separately. After the integration, Acrobat will become the recommended app in Gmail and Google Drive and will appear in the Edit with item in the file preview. According to Adobe, more than 400 billion PDF files are opened in Acrobat every year.
Acrobat's AI features will also run on models from the Gemini family. These include editing a PDF in a conversational format, analyzing a document's contents and preparing presentations, summaries and reports based on it. Adobe will also expand its connectors in Gemini and Gemini Enterprise: Acrobat and additional creative tools will be added there. The company estimates the audience of Adobe's creative and office products at more than 1 billion monthly active users.
What is missing from the announcement
The main practical question remains open. The companies did not say how many generative credits Veo, Nano Banana, Gemini Omni and the new Google models will cost in Firefly and Photoshop. Nor is it known which of them will be unlimited on paid plans. For a studio generating video in batches, the budget depends on this more than on the fact of integration itself. It has also not been disclosed how artists' files and clients' documents will be processed when Acrobat works inside Gmail and Drive. There is no data on plan and regional restrictions either. Agencies with strict confidentiality agreements would be wise to wait for these details.
Judging by the other MAX announcements, Adobe is betting on openness to other companies' AI. As TechCrunch notes, the company released a solution based on the MCP protocol through which third-party chatbots, including ChatGPT, Claude and Gemini, can use Adobe's features. In the same piece, the publication writes about the Builder prototyping tool in Firefly, which has entered closed beta, and considers it an incursion into Figma's territory. The partnership with Google apparently fits the same logic: Adobe gathers external generators in its apps alongside its own models and brings its tools to where users already work. For an artist this means a choice of models inside the familiar Photoshop. The flip side is that the quality and price of the result depend on the terms Adobe agrees with Google.
Why is Adobe increasingly embedding Google's models into Photoshop: the lag of its own models, pressure from competitors, a chase for Google's audience?