Generative AI startup Leonardo.Ai has been acquired by Canva. Founder JJ Fiasson explains what happens next.

The founder of Leonardo.Ai says Canva 's decision to buy his generative AI startup is based on the "synergies" between the pair, and not because the Australian tech giant saw it as a threat.

"Canva has a very strong internal generative AI team,' says JJ Fiasson , the founder of tells LinkedIn News Australia.

"This is about augmentation and supercharging what's already there. It also allows us to gain more resources to scale our product and research efforts."

Canva, which is expected to go public in 2025 or 2026, has been shaping itself as a rival to Adobe through recent acquisitions, as it aspires to build a “world-class suite of visual AI tools”.

In March it also bought Serif, which makes the Affinity creative software used by millions of professional designers around the world.

Read the full interview with Fiasson below to learn what's in store for Leonardo.Ai once it becomes part of the Canva family, and for his insights into the impact generative AI is having on art and culture.

JJ, we will get to the Canva news in a moment. But for those who are new to your company, can you explain where the idea for Leonardo.Ai came from?

I previously co-founded a gaming studio where we were looking at ways to increase the efficiency around visual content creation. I went down a rabbit hole playing with things like Disco Diffusion, which was an early diffusion model where you could use a text prompt and generate an image. It would take about 20 minutes to generate something and it didn't look very good, but it was the promise of things to come.

As the technology progressed, I could really see the huge potential for generative AI, especially in the visual space, to revolutionise content creation.

So I put together a team. I met a couple of very passionate AI researchers who had been contributing to an open-source project and we formed this founding team motivated by the aim of bringing control and consistency to the equation. We felt that was key to broadening productive use cases and driving further adoption

That has been our approach from the beginning — keeping the user in the driver's seat making people feel in control of the creative process, while being augmented by generative AI.

What’s your point of difference as an image generating platform?

We are one of the few platforms that actually has an expansive toolkit for creators, in terms of the number of ways that you can interact with the models on our platform to feel like you have more control.

It's not just a text prompt. You can also use a real time canvas, for example, to draw and have your rough sketch augmented in real time into something far more polished. There's a whole range of different inputs.

There's this level of control and consistency on the platform that is hard to find in other places. Plus, we have recently launched our own foundational model, Phoenix, which is an extremely powerful prompt adherent model great at doing accurate text rendering within images.

Was it daunting coming to the generative AI race when you formed in 2022? There were already companies like Midjourney and OpenAI well ahead.

I don't think that we were late to the race, to be honest. We started in October 2022, which was actually early in the cycle. I don't think the ChatGPT [demo] had even been released at that point. We launched after the initial release of Stability AI's Stable Diffusion.

I think the larger players in the space generally have been more focused on the deep tech questions, not necessarily on the application layer side of things. Our approach was to try and work on both and lean a lot into open-source as well. We've integrated open-source models into the platform. We've also contributed to open-source as part of our own research work.

The Canva deal is huge news. How did it come about?

We are both Blackbird portfolio companies and have been chatting with the Canva team on and off for a while. I think they've been very interested in what we've been doing.

We weren't seeking to be acquired. We were actually planning our next fundraising round. However, the opportunity came along and it made a lot of sense from our perspective.

We are big admirers of everything that Canva has achieved to date and they've got a very similar set of values and goals in terms of empowering the world to design and create.

It's just ultimately a great match-up in terms of the synergies between the two companies. It also allows us to gain more resources to scale our product and research efforts.

Canva Co-Founder Cameron Adams said in a LinkedIn post that "uniting our teams will accelerate the work we’re each doing in research and development". Does being part of a big tech company change Leonardo.Ai's original goals and mission?

I don't think that our mission is actually going to be very different to be honest. Canva is very supportive. Firstly, we are continuing to remain as an independent structure within the Canva family and Leonardo will continue to build our own platform, while scaling our product and research teams.

We will also be looking at integrations into the Canva platform and supporting them with our research work as we continue to build our own platform, effectively gaining huge distribution through the Canva platform. We aren’t making any major shifts from the direction we were already going. This allows us to think potentially bigger than we would have otherwise.

Can you disclose the price tag of the deal?

I’m not going to disclose the finances of the deal.

Did Canva acquire Leonardi.Ai because it saw it as a challenge or threat?

I don't believe that to be the case. Canva has a very strong internal generative AI team, and they've already done some really awesome stuff. I think this is about augmentation and supercharging what's already there.

They’ve also seen our achievements as a company in terms of our rapid growth. I think they really appreciate the product that we've built and how powerful it is.

It’s also different in the way that someone might create on Canva and appeals to a different subset of users.

And a lot more of them. What will it mean for you having access to Canva’s monthly activer user base of 190 million people?

That's a great opportunity for us. Leonardo.Ai already has roughly 3 to 4 million active users a month. For us to be able to have integrations into Canva and to its giant user base is really exciting. We love building cool stuff. Having more people use that cool stuff is an exciting prospect for the team.

How will you integrate into Canva?

That's currently something that we are working through at the moment. It has not yet been decided.

Does this partnership mean Australia could soon have a world-leading generative AI company to challenge some of the global players like ChatGPT and Midjourney?

Canva has already done some awesome stuff with respect to generative AI and we have made our mark as well. I definitely think the two companies joining forces will take that to new levels.

I believe having different ways of leveraging generative AI is useful.

There's this ‘democratisation’ element in terms of leveraging generative AI within larger platforms like Canva, especially platforms that have a generous ‘freemium’ model.

Is it ethical to train AI on artists work, and can we considered the output 'art'? I’m interested in your view on this given your own creative background.

I can definitely see different sides of that argument.

Ultimately AI models are a reflection of society and culture.

I think it's important that there is a level of understanding that models have to adequately represent the culture and society that they're seeking to reflect.

In terms of how we think about data and models learning and gaining a view of the world, I think there's a lot to be said for contextual understanding and having a good comprehension of art history. If you think about how a human learns, they learn by going to art galleries, to museums and having a really strong exposure to collective human culture and history.

It’s very legitimate for people to feel and to have an issue when AI models are used to recreate, existing works of art by artists or recreating someone else's style. That’s not a great use of the technology.  I think that kind of shows a lack of imagination.

The use of genAI for art needs to be in the context of novel creations. What's also really important and what we like to focus on, is how people think about the intent — the use and the outputs of a model. What are people using these platforms for? When people are using them to create novel works of art and, for practical everyday use cases — and that is by far the bulk of what we see — I think a defensible use of the technology.

This technology is ultimately something that can augment human creativity.

Should AI imagery and art be its own separate category then? In the same way photography isn’t considered painting, but is still a form of expression.

If art has a human with intent behind the creation, then I think you can definitely make a case for something to be considered art. I think it ultimately depends on the level of human thought and intent.

Historically, if we look at the use of technology throughout history and look at artists like Vermeer, it was long thought that his use of the camera obscura to be able to better capture colour and light was an example of using technology.

Fast forward to the camera, that was something that people saw as the antithesis to artistic creation at the time and it changed the way the portraiture industry worked. I think similarly, when computers came about and people were able to use Photoshop, that was something that changed how people thought about artistic creation.

I think generative AI is ultimately a product of collective human creative experience because it reflects culture and everything else it's been exposed to. Ultimately, that is something that can augment human creativity. If you are someone who is creating using some of these tools, you can say that the output is art because there is artistic intent there.

Do AI companies and big tech companies have a responsibility in ensuring generative AI content is not misleading? And what does Leonardo AI do to ensure its users aren’t creative misleading content?

It’s definitely something contingent on the major platforms as to what they can do to limit deep fakes or false information. That is something that we, and the entire industry, take very seriously.

We have a number of guardrails on our platform around that. We have a range of filtering mechanisms that detect on the input side what the user is asking for and to try and identify if there's malintent on behalf of the user, and blocking at that point. There is also filtering on the output side.

💡 Want to learn more about generative AI, digital design and content creation? Follow Leonardo.Ai's JJ Fiasson for more insight. You can also follow Canva Co-Founder Cameron Adams .

➕ Click here to subscribe to the latest tech and startup news and analysis on LinkedIn.