Adobe Firefly Updates Provide Insight into Adobe’s AI-Native Platform

Adobe has been silently investing in generative AI capabilities even before the 2022-2023 AI boom. Over the past decade, experiments with AI frameworks, including Adobe Sensei, paved the way for automation tools and features within Adobe Acrobat and other apps across the software giant’s Creative Cloud suite.

Today, Adobe’s proprietary AI platform, Firefly, is seeking to expand its commercial use cases in generative AI for enterprises, small businesses, and individual creators. This shift is particularly relevant for Australia’s creative economy, where agencies, in-house brand teams, and independent creators increasingly operate across global markets while needing strong local compliance and IP protections. Following Adobe MAX to close out 2025, the year’s first batch of Firefly updates has provided insight into Adobe’s vision for the future of AI, both within and outside of its own software offerings.

Here’s how Adobe Firefly updates are positioned alongside other AI tools expected to shape workflows through 2026, including for Australian creative teams navigating tighter budgets, faster turnaround expectations, and global clients.

‘All-in-one’ creative AI capabilities

Adobe MAX highlighted the software giant’s strategic shift towards transforming Firefly into an ‘AI-native’ platform. This means that, alongside being a top-performing AI image generator, Adobe Firefly also integrates generative AI across audio and video outputs. 

Now with Adobe Firefly’s AI tools, users can:

• Generate images from text prompts
• Generate sketches from images
• Generate videos from images and/or text prompts
• Extend videos based on clipped video prompts
• Translate audio and video into over 20+ languages

For Australian agencies working with international brands, particularly across Asia-Pacific and Europe, these multilingual and multi-format capabilities reduce the need for external localisation workflows. The expanded output of Adobe Firefly to include a wider variety of image, audio, and video generation capabilities positions the tool to offer creative support from the brainstorming phase and all the way through to final production. 

Firefly’s ‘Prompt to Edit’ features also give users greater control over the polishing and refinement of their generated outputs, simply by providing the tool with natural language instructions, a practical advantage for small Australian studios and freelancers without large post-production teams.

Firefly users of today can effectively integrate this AI tool over a greater number of projects, which Adobe states makes Firefly a platform being evaluated by creative industries and enterprise users.

Expanding closed LLM training data

Adobe continues to maintain its focus on quality of output by updating Firefly’s training databases, which the company states are also entirely self-managed with zero dependence on publicly available LLM training datasets. The investment in closed databases for LLM training works to reduce risks of both AI hallucinations and copyright infringements.

For Australian businesses operating under local IP laws and working with global clients, this focus on ownership and provenance is particularly significant. Adobe emphasises that users retain full ownership rights over Firefly-generated outputs, supported by its ‘Content Credentials’ system, metadata that identifies AI-generated content and its creator.

Commercial safety and trust

Adobe reinforces that Firefly AI models are trained on both openly licensed and public domain content, as well as Adobe Stock, the company’s internally owned and operated stock assets library. Maintaining ethical practices in training their AI models is foundational to supporting Firefly’s positioning as a ‘commercially safe’ AI.

Maintaining close LLM training databases is just one piece of the equation when it comes to establishing Firefly as a ‘commercially safe’ AI-native platform. To keep Firefly suitable for commercial users, Adobe also invests heavily in quality control measures that work to maintain consistency in output for Firefly tools. Users with proficiency in AI prompting can effectively get the output they’re after faster and easier when using Firefly in comparison to other generative AI tools. For Australian enterprises in sectors such as retail, media, education, and government, where brand consistency and compliance matter, this reliability is a key differentiator.

Adobe also publishes prompt libraries and educational resources to support user proficiency, resources increasingly adopted by Australian creative teams building internal AI guidelines and workflows.

Agentic AI and workflow automation

While the past few years of Firefly updates have been all about establishing Adobe’s proprietary AI tools as industry leaders similar to other large-scale AI platforms, the most recent platform updates following Adobe MAX also signal Firefly’s move towards agentic AI and workflow automation.

There are three key updates in the realm of agentic AI and automation features that are worth exploring in Adobe’s most recent batch of updates:

1. Project Moonlight

Further integrating Firefly into the wider Creative Cloud suite is essential to this next phase of the platform’s development in an AI-forward business landscape. Adobe’s getting started here with the introduction of Project Moonlight, their new AI agent, or as Adobe refers to it, ‘your personal orchestration assistant’. 

Project Moonlight basically coordinates Firefly outputs across Adobe’s other apps – from Acrobat to Photoshop, and everything in between. Programmed to provide personalised recommendations and workflow assistance, Project Moonlight can be used to take the manual admin out of switching from platform to platform when working across dynamic, multidimensional creative projects.

2. Firefly Creative Production

A unique editing feature that definitely reflects Adobe’s enterprise inclusion, Firefly Creative Production basically supports batch editing across large image sets. For users working with a large batch of image assets that all require the same edit effects, Firefly Creative Production can be used to automate the process. 

The automated editing processes here can also be as detailed or as simplistic as needed. Just looking to remove image backgrounds across 300 photos? Sorted. And if you have filter applications or colour grading to add to that workflow, Firefly Creative Production can accommodate these edit settings as well.

3. Firefly Boards

The tool lowers the friction typically involved in revising and reworking ideas. With Firefly Boards, users can brainstorm directly within their AI-native workspace and take concepts from rough drafts to polished assets faster and with greater focus. 

With Firefly Boards, users can also access some unique tools, including options to rotate two-dimensional images into three-dimensional renderings and export designs to a .pdf file format for easy sharing.

Firefly integration for the Creative Cloud suite

Another element of the transition from traditional generative AI to agentic models is further supporting Firefly integration across Adobe’s wider Creative Cloud suite of tools. Now, Creative Cloud users can enjoy all the advantages of Firefly Image Model 5, delivering improved photorealism, anatomic accuracy, and native 4MP resolution for print-ready design outputs.

Updates to the Adobe iOS and Android Mobile apps also make it easier for users to access Firefly AI features while editing on the go. So whether you’re accessing the Creative Cloud tools on your desktop, in your browser, or via your smartphone, users will encounter a similar set of high-capacity AI features, which is particularly useful for Australian creators working on location, across events, or in hybrid environments.

Hybrid model ecosystem

Alongside improving Firefly integration within the Creative Cloud suite, Adobe is also investing in partner model integration for users who are working across other industry-grade AI platforms. As of October 2025, Adobe Firefly now integrates third-party AI models, including models from:

• Runway (Aleph)
• Topaz Labs (Astra)
• ElevenLabs
• Google (Veo)
• Black Forest Labs (FLUX.1)

Firefly Foundry updates

Finally, the release of Adobe Firefly Foundry has also been a major bit of news from Adobe. The new platform, leveraging AI for commercial-grade media production, has reported adoption among enterprise organizations in sectors such as consumer goods, entertainment, and consulting. 

On top of providing account features that are ideal for brands, like the ability to create custom templates, Firefly Foundry also allows commercial users to develop their own proprietary, IP-protected generative AI models that are trained directly on a user’s brand resources. 

Firefly Foundry models are also available across the full Adobe ecosystem, including GenStudio, Creative Cloud, and Express, which may allow commercial users to generate.

For Australian brands and agencies, Firefly Foundry offers the ability to train proprietary AI models on local brand assets, which is an increasingly attractive option for organisations seeking to protect IP while scaling content output.

What’s next for Adobe’s ‘commercially safe’ AI?

Looking ahead to 2026, Adobe is expected to continue expanding Firefly’s integrations across both Adobe and third-party platforms. For Australian businesses balancing innovation with regulation, Firefly’s emphasis on commercial safety, ownership, and workflow efficiency positions it as a practical AI platform rather than an experimental one.

As Adobe continues to evolve Firefly based on user needs, Australian creators and enterprises are likely to play a growing role in shaping how AI is applied responsibly at scale.

Stay up-to-date with Adobe news within the platform.