Google on February 1 updated Bard, giving the generative AI chatbot the ability to generate images based on user prompts. This will be powered by Google’s Imagen 2 text-to-image model.
With this feature, Bard catches up with its rival ChatGPT, which has been offering image generation capabilities since September 2023 thanks to the integration with OpenAI’s DALL-E 3. However, ChatGPT only offers this feature to its paid users, whereas Google is offering it to all users for free. But from our use of the two, it appears that ChatGPT has a leg up in terms of the quality of the images generated.
This new capability comes at a time when AI-based image generation is facing scrutiny and criticism owing to the rise in deepfakes, especially the recent incident involving sexually explicit deepfakes of Taylor Swift. While tools like Bard and ChatGPT are not commonly used for deepfake generation, they are not immune to other forms of unethical use.
Google said that it has “added technical guardrails to limit problematic outputs like violent, offensive, or sexually explicit content as well as applying filters to reduce the risk of generating images of named individuals.
Additionally, all images generated by Bard will be marked by SynthID, a watermarking tool developed by Google that allows anyone to detect AI-generated images. You can read more about SynthID here.
Google also released another experimental image generation tool called ImageFX, which is powered by Imagen 2 as well. It’s not yet clear what is the difference between image generation on Bard vs. ImageFX, but it appears that the latter will have more features specific to image generation.
Currently, image generation in Bard is only available for English prompts and for users above 18. Also, it is not yet available in the European Economic Area (EEA), Switzerland, and the UK, likely due to the stricter regulations in these regions.
- Google Launches Watermarking Tool For AI-Generated Images
- Taylor Swift Deepfakes Bring Increased Attention To AI Regulation
- DEFIANCE Act: New Bill Proposed In The US To Curb Non-Consensual AI Porn And Deepfakes
- Explainer: Why Detecting Deepfakes Is A Challenging Problem #NAMA
STAY ON TOP OF TECH NEWS: Our daily newsletter with the top story of the day from MediaNama, delivered to your inbox before 9 AM. Click here to sign up today!