The news: Advertising agency Batten, Barton, Durstine & Osborn (BBDO), the largest in ad giant Omnicom Group’s portfolio, sent a memo to employees this week warning them against using generative AI like ChatGPT, MidJourney, and DALL-E for fear of copyright violations and other concerns.

The dos and don'ts: BBDO CEO Andrew Robertson told employees they shouldn’t use AI for client work unless it’s approved by the agency’s legal team; he also advised speaking to leadership before discussing AI with clients. Some BBDO clients have already rejected work that used generative AI, according to Robertson.

• The main reason for concern is that generative AI tools are often trained on a broad range of material, including copyrighted content. Several major lawsuits against AI companies allege that constitutes copyright infringement, and the US Copyright Office recently said that most AI-generated works can’t be copyrighted and should be reviewed on a case-by-case basis.
• But still, BBDO is encouraging its employees to experiment with the tech and find potential use cases—just not in the public eye, and certainly not to create work that could land clients in legal trouble. Like many agencies, BBDO uses AI for data analysis and purchasing purposes.

The rift: Advertisers have been uncertain about generative AI’s legal status since the tech burst on the scene, though they’ve still experimented behind the scenes. But BBDO’s warning stands in stark contrast to the attitude of ad giants Google and Meta.

Despite legal concerns, Both Google and Meta have announced plans to launch generative AI tools for ad campaigns later this year. While that will almost certainly create a huge wave of AI-generated ads, larger advertisers may still choose to stay away. Such products may be best suited for small businesses that can’t afford to outsource creative work—but who also probably can’t afford a costly legal challenge.