Meta rolls out AI-driven brand safety and verification tools

Meta launched new inventory filters for Facebook and Instagram for advertisers in English- and Spanish-speaking markets, according to information shared by the company. The tools use a multi-stage AI review system to classify text, photo and video content and are intended to provide more control over the proximity of ads to potentially objectionable content.

The controls define high-, medium- and low-risk content. Advertisers can choose from three settings to control the types of content that appear above and below an ad.

Meta has also partnered with data and technology provider Zefr to create an AI-powered tool that can assess video, image, text and audio to report the context in which ads appear on a Facebook Feed. The new tools show Meta looking to shore up advertiser concerns by providing new ways to help classify content and context.

Meta, the parent behind Facebook and Instagram — still the gorillas in the social media room in terms of mass and usage — is looking to shore up advertiser concerns (and distance itself from the issues plaguing other social media companies) with several new tools. While social media platforms as a whole have been demonstrating improvement in the areas of safety, inclusivity, sustainability and data ethics, the channel is still a long way away from the trust and controls that are already in place for traditional broadcast and cable networks, according to IPG Mediabrands and Magna’s Media Responsibility index .

The new controls will evaluate if the content meets the company’s exist ing monetization policies. If it does not, the content will not be eligible to have ads above or below it. Content that does fit the guidelines, however, will be classified into high, medium and low-risk categories based on Global Alliance for Responsible Media’s (GARM) suitability framework. Advertisers will then have the opportunity to define their content preference levels. The options are:

• Expanded inventory, which will allow ads to appear next to all content that meets both Meta’s Community Standards and its monetization eligibility criteria.
• Moderate inventory, which will exclude high-risk content.
• limit ed inventory, which will exclude both high- and medium-risk content.

The controls are only currently available to advertisers in Meta’s English- and Spanish-speaking markets. The company will expand the controls to support additional languages and markets later in the year. Meta plans to test the tools on Reels, Stories and Video Feeds as the company learns more about advertiser preferences.

Content is one thing, but context also matters. To that end, Meta has been working with Zefr to develop an AI-powered tool that will evaluate the context in which ads appear on a Facebook Feed. By assessing and labeling video, images, text and audio content based on GARM’s standards, the tool will help advertisers measure, verify and understand the suitability of content near their ads.

That risk may be low, however. Meta says testing indicates that less than 1% of the content evaluated falls into the high-risk category. The verification tool will be rolled out to badged Meta Business Partners later this year.

Finally, Meta emphasized its exist ing Community Standards and Guideline enforcement practices, such as publishing its quarterly Community Standards Enforcement Reports (which have been independently assessed by EY) and its recent accreditation from the Media Ratings Council (MRC) for its Partner Monetization Policies, Content Monetization Policies and content-level brand safety and suitability controls for In-Stream Video.

“We are committed to giving advertisers more control and transparency and will continue ensuring our work has independent oversight,” said Samantha Stetson, Meta’s vice president of client council and industry trade relations, in a post on the company’s website.