Adobe's creative AI agents now span Photoshop to Premiere

Disney and Adobe are partnering to build custom generative AI models tailored to Disney's creative and brand requirements. The collaboration centers on Adobe's Firefly platform, which is designed to produce AI-generated content trained on licensed material. The goal is to give Disney's production teams AI tools that stay within the strict visual and narrative boundaries the company maintains across its properties.

ByteDance has released Seedream 5.5, an image generation model aimed at professional production use. It promises pixel-stable editing across up to 100 iterative passes without color drift, and adds native RGBA output with transparent backgrounds. The company also claims it outperforms GPT-Image-2.5 Sunburst on realism benchmarks.

At its Max conference in Miami, Adobe outlined a broad push toward agentic AI, positioning chatbots and autonomous agents as potential collaborators in creative workflows. New features span Firefly, Creative Cloud, and third-party integrations - pointing to a future where repetitive design tasks can run largely on their own. The announcements signal a meaningful shift in how Adobe imagines its tools being used day to day.

At Adobe MAX 2026 in Miami Beach, Adobe unveiled a set of AI-powered updates to Photoshop and Lightroom, including tools that can relight and restyle existing photos. The additions reflect Adobe's ongoing effort to fold generative capabilities directly into its established creative workflows. This article, reported by PetaPixel, covers the key announcements from the event.

Adobe has pushed a significant update to Premiere Pro, moving its Color Mode feature out of beta and bundling in a range of new generative video and AI-driven audio tools. The release reflects Adobe's continued effort to weave machine learning assistance throughout its professional video editing workflow. Here is what editors can expect from the latest version.

Adobe announced at its MAX 2026 conference in Miami that local MCP (Model Context Protocol) integrations are coming to Creative Cloud applications, including Premiere, After Effects, Photoshop, and Illustrator. The feature is expected to reach Creative Cloud customers in early 2027, giving editors and creators the ability to control native project files through third-party AI agents. Layers, vectors, bins, and sequences are set to remain fully editable throughout these agent-driven workflows.

Google has introduced the Multimodal Knowledge Agent (MKA) benchmark, a new evaluation framework designed to test how AI agents reason through visually complex, multimodal tasks. Unlike simpler benchmarks, MKA focuses on the kind of layered reasoning required when an agent must draw on both visual and textual information simultaneously. The release signals growing interest in more rigorous, real-world testing standards for multimodal AI systems.