AI Safety at Canva

How we safeguard our community

At Canva, our values play an incredibly important role in guiding how we build our products, treat our community, and the impact we want to have on the world. We’re committed to building a safe and trusted environment where our community can unlock their creativity and bring their ideas to life. To achieve this, we’ve implemented multiple layers of safety measures in our Magic AI-powered products, including but not limited to:

• Reviewing the input - We automatically moderate certain input prompts to identify and prevent things that might generate inappropriate content.
• Reviewing the output - We automatically moderate generated content when we determine the result might be unwanted or inappropriate.
• Reducing bias - AI systems and datasets can promote or reduce unfair biases, and concepts of fairness can vary across different cultures. We take steps to minimize unfair bias and misrepresentation.

We also use a range of tools to help to keep our platform safe, including machine learning. For example, our tools help our community adhere to our AI Product Terms by aiming to prevent the creation of the following kinds of content:

• Medical topics, such as medical advice and information.
• Self harm and mental health topics. We encourage contacting local mental health support services.
• Sexually explicit content.
• Political topics, including content about politicians and elections.

As always, we’re constantly looking at ways to continue building the world’s most trusted platform. If you see anything inappropriate or unexpected, or feel that your prompt has been incorrectly flagged, please submit a report so we can look into it.

• When using Text to Image, you can report inappropriate content using the **Report** button in the app.
• When seeing incorrect flags in Text to Image, or using any other AI-powered products, such as Magic Write™, Magic Edit, and Translate, you can report inappropriate content or incorrect flagshere⁠(opens in a new tab or window).