A month ago, Meta unveiled a set of generative AI tools for advertisers. “We believe that these features will unlock a new era of creativity that maximizes productivity, personalization and performance for all advertisers,” enthused monetization infrastructure VP Matt Steiner at the time.
Well, not all advertisers, as it turns out. The social giant is now banning the tools’ use in making ads related to “housing, employment or credit or social issues, elections, or politics, or related to health, pharmaceuticals or financial services.” This is so Meta can work on building “the right safeguards for the use of generative AI in ads that relate to potentially sensitive topics in regulated industries.”
We’re probably going to see more sharp course corrections of this sort, as tech conglomerates start to spot problems churned up by their frantic race for AI relevance. I’m not sure why the proprietor of Facebook and Instagram, which are constantly struggling to rein in political disinformation, took a moment to figure out that this task would not be made easier by giving everyone tools to generate convincing fakes—but here we are. I can only imagine the cognitive dissonance going on within Google, which is giving people the means to generate faux copy while also trying to maintain a search engine that points users to useful, original content.
Google, of course, has also offered advertisers a set of genAI tools. And like Meta, it’s trying to avoid their use by propagandists—per Reuters, a list of “political keywords” will be banned as prompts, and election-related ads will have to disclose “synthetic content that inauthentically depicts real or realistic-looking people or events.” Good luck enforcing that in the massive election year of 2024, if the enormous progress made by image generators in the last 12 months is anything to go by.
Meanwhile, Meta and Google will soon face stricter limits on their political advertising businesses in Europe—though perhaps not soon enough.
Next year, voters will usher in a new European Parliament and possibly new governments in Austria, Belgium, Croatia, Lithuania, and Romania. These will be the last of these elections that won’t be covered by a new EU regulation on the transparency and targeting of political advertising, which is currently nearing the end of the EU’s legislative process—EU lawmakers and national representatives announced today that they have reached a provisional agreement on the final version.
The law will ban the use of personal data for online political-ad targeting unless the data comes straight from the data subject, who must give “explicit and separate consent” for this use. The use of “data revealing racial or ethnic origin or political opinions” to target political ads will be strictly forbidden. All this will apply 18 months after the bill becomes law, so let's just say European elections taking place from late 2025 won’t be such big money-spinners.
More news below.
David Meyer
Want to send thoughts or suggestions to Data Sheet? Drop a line here.
