Google's Bard can finally generate images

For months, OpenAI held a massive trump card over Google: if you wanted a top-tier chatbot that could also draw whatever you asked for, ChatGPT Plus with DALL-E 3 was basically the only game in town. But that perk came with a $20 monthly subscription fee.

Now, Google is finally striking back. The tech giant just rolled out AI image generation directly inside its Bard chatbot, letting users create pictures at no cost.

Imagen 2 meets Bard

The new capability is powered by Google’s Imagen 2 text-to-image model. While industry observers previously expected photo generation to arrive alongside the still-in-development Gemini Ultra, Google decided not to wait, deploying the feature into the current Gemini Pro-powered Bard.

Users can simply type a text prompt to generate images directly in the chat interface. Unlike ChatGPT Plus, which keeps DALL-E 3 locked behind a paywall, Bard is offering image creation for free.

Google is also rolling out ImageFX, a standalone experimental image-generation tool inside Google Labs. Much like MusicFX and TextFX before it, ImageFX is designed for users to play with generative AI and send direct feedback to Google engineers.

Safety is a glaring focal point for this rollout. Following recent controversies where explicit fake images of Taylor Swift flooded X, Google emphasized that Bard’s image tools were built with strict guardrails. The system is programmed to avoid generating images of well-known individuals and block violent or sexually explicit outputs.

Every image produced by Bard and ImageFX will also feature Google's SynthID watermark baked into the pixels, making it machine-detectable as AI-generated media.

Alongside image generation, Google significantly expanded Bard's reach. While image prompts must be in English for now, the Gemini Pro update has expanded from 170 countries to over 230, supporting more than 40 languages including Arabic, Bengali, Tamil, and Urdu. Its double-check verification tool is also now live across those 40 languages.

Why it matters

Google making high-quality image generation completely free is a big win for everyday users and casual creators. You no longer need to shell out a monthly subscription to brainstorm visuals, mock up concepts, or generate illustrations.

We think this is a smart tactical move by Google. Instead of holding back until Gemini Ultra is ready, pairing Imagen 2 with Gemini Pro immediately closes Bard’s biggest multimodal gap against OpenAI. By pricing it at zero, Google puts real pressure on ChatGPT Plus retention among non-power users.

However, there are trade-offs. The strict guardrails against generating well-known people will protect Google from public relations disasters, but overly cautious guardrails often frustrate creators looking for flexibility. Furthermore, keeping image prompts English-only blunts its global launch until multilingual generation arrives.