Meta is giving advertisers new generative AI tools in its ad platform, a development the company started promising this year as AI became a bigger part of the discussion in marketing with the rise of ChatGPT.

Meanwhile, this week Google also released a host of new AI capabilities across products, including search, which will have drastic implications for advertisers.

Meta unveiled a program today called AI Sandbox, which is a testing ground for brands to apply generative AI to text and images in campaigns that appear in feeds and videos on Facebook and Instagram. So far, AI can help automate some image touchups, including switching backgrounds on ads that contain pictures of products, but Meta executives envision more sophisticated uses. AI-generated text will be used to personalize messages in ads, too.

“AI is going to be able to help with the creative process,” Nicola Mendelsohn, Meta’s head of global business group, said at a press event today in New York to explain the AI functions.

Meta did not say how many advertisers were applying generative AI to their campaigns, describing it as a “small group,” with plans to roll out to more advertisers in July. Generative AI is not yet touching videos, but Meta is exploring that area, according to John Hegeman, Meta’s VP of monetization, who also spoke at the event.

To demonstrate the possibilities, Meta pointed to Jones Road Beauty as one of the brands playing with generative AI. The brand used AI Sandbox tools to produce text in ads and images for the backgrounds. The tool works similarly to the generative AI platforms that have become widely available, such as ChatGPT and DALL-E, where the user issues a prompt and the machine provides a result. Meta's AI also helps crop ads so that they could run in feeds that require one size, or within Reels and Stories videos that require a different orientation.

More news: How OpenAI is working with brands

Meta’s generative AI is an evolution of an automated ad product called Meta Advantage+, which specializes in driving sales and other outcomes from campaigns. Meta has been using AI in the ad platform for years to help with changes to how data is shared online, after the pending death of cookies and Apple’s moves to block data sharing. AI is doing more of the targeting and measuring in ads.

With the interest in generative AI, after ChatGPT entered the mainstream with a consumer-facing chatbot, advertisers started looking for ways to put the creative tools into building ads. Microsoft, Snapchat, Google, Meta and TikTok are all exploring ways that AI can fit into products and services. This week, at Google I/O, the developer conference, Google CEO Sundar Pichai talked about integrating AI into 25 products and features. Google also made Bard, its search-based chatbot, more accessible, giving people a way to search the web using more natural language-based queries. 

Meta has been talking about chat features that could help businesses, too. Meta has click-to-message ads in its apps, and while those do not use the latest generative AI to produce conversations with consumers, that is the plan, Hegeman said.

Meta executives said the company was still working out how it will disclose when AI is used in ads. One of the concerns about AI-generated content is that it can be deceptive, depicting scenes and people that don’t exist. The first products in Meta’s AI Sandbox don’t generate any images of people and don’t make videos.