Insights
User Consent and Meta AI
User Consent and Meta AI: What it could mean for brands
Meta has been in the news lately for announcing that, starting June 26, individuals in the EU and UK will have the option to opt out of Meta AI using their content to train the model. This is significant progress for data privacy in Europe. While the U.S. isn’t there yet, similar measures are being considered with the AI Consent Act. Currently, users in the U.S. can’t opt out of Meta AI training, but their private messages, according to Meta, will not be used to train the model.
This step towards greater consent is something many have been hoping for from AI platforms. But what impacts could this have on brands’ use of Meta AI?
A Potential Decrease in Generative Quality
Generative AI relies on huge amounts of data to improve its accuracy and creativity. If a significant number of users opt out or turn to private messages, Meta AI might not have the data to help brands create great, personalized content. Brands could be slow to adopt.
In addition, will certain audiences be more likely to take these protective steps? Younger adults tend to be more positive about AI than older adults and may be less likely to make a change in their behavior. If demographic groups are unevenly represented, brands should be prepared for an increased bias in Meta AI’s generative outputs (an area already of concern with generative AI).
Reduced Audience Information
Audience insights are vital for targeted advertising and campaign planning. If fewer users are sharing data, the depth and accuracy of audience segmentation might not be as strong as they’re led to believe. Brands may need to invest in alternative data sources or platforms if they feel the audience insights from Meta are lacking.
Depending on how audiences react to and take steps to protect their privacy from Meta AI, brands could also face challenges in crafting targeted messages, which is one of the major promises of generative AI.
Brand Content and Data Protection
Brand content isn’t protected from Meta AI training, so brands should identify where, when, and how they use Meta AI in their processes.
A responsible brand should scrub posts and ads for sensitive information, so this aspect would seem moot. But the data a brand inputs when prompting Meta AI is where things get tricky. A brand might use Meta AI to write headline copy and add sensitive brand audience data to the prompt to help the AI arrive at better outputs. Because the brand can’t opt out, this data becomes part of the model and can be used by the model to help competitors.
Ethics and Brand Reputation
Trust is a major contributor to brand success, so brands should consider how their audiences will react if they use Meta AI. Depending on audience sensibilities around data privacy, brands may want to be transparent about their ethical practices with AI. Brands that prioritize privacy and consent could even see an increase in their reputation.
Should Brands use Meta AI?
In the age of AI, we believe consent is more important than ever, regardless of the allure of AI’s “better, faster, cheaper” promise. We applaud Meta for taking steps to allow individuals to protect their content. But that doesn’t mean we’ll be using Meta AI as one of our generative tools any time soon. When using client data with generative AI’s, we only use tools that allow us to opt out of training the model for everyone or specifically state that they don’t train the model on user data. We recommend any brand or agency to do the same.
That said, the rapid pace of change with all things AI makes it hard to have a definitive, long-term stance. In addition, there are multiple use cases to consider besides using Meta AI for produced, audience-facing work (which has been the focus here).
Ultimately, brands need to take the time to define their values with respect to AI tools and have a nimble plan in place to adjust as needed with the rapid changes. Ethics should be the guide, not the glitter of innovation.
Additional resources:
Facebook Privacy Policy
Privacy Matters: Meta’s Generative AI Features
AI Consent Act
 |
Adam Deer
|