A Shifting Onus: What Meta's Content Moderation Changes Mean for Advertisers | Brand Safety Institute January 2025

On Tuesday, Meta announced a sweeping set of changes to its content moderation policies and processes, including the company’s decisions to end its third-party fact-checking program, lift restrictions on several content areas, relocate its policy and enforcement teams, and revisit its approach to political content. These changes are not surprising in light of recent election victories for the political right in the U.S. and around the world. While the company remains focused on ensuring brand safety & suitability, these changes further galvanize an industry-wide pivot towards individual choice, shifting the responsibility for decision making around content moderation and monetization away from platforms and towards individual users and advertisers.

I’m not here to argue whether these changes are right or wrong, but having spent nearly seven years leading brand safety efforts at the company formerly known as Twitter, and having had to explain and defend many of these same changes publicly in the past, it felt worth sharing my thoughts on Meta’s announcement and what it means for advertisers. I’ll also share a set of action items for advertisers navigating these changes.

Moving from Third-Party Fact Checking to Community Notes

First, Meta announced that it is ending its third-party fact-checking program (starting in the U.S.), opting instead for a Community Notes-based approach and explicitly citing X’s decentralized, user-based fact-checking program of the same name as a model for this strategy.

This move reflects several realities. First, it’s really hard to write and enforce misinformation policies as a platform that appease everyone. Historically, platform policies around disputed information have centered on a few high-impact areas, such as vaccines and elections. Different sides have argued that these approaches are either too narrow or too restrictive, arguing respectively that platforms should write and enforce broad policies for misinformation on any topic, or that platforms are either biased in their enforcement or that they shouldn’t be arbiters of truth.

Decentralized solutions such as Community Notes (which, for the record, were in development at Twitter for nearly two years before Elon Musk acquired the company) have the benefit of allowing users to add context on any subject, while also reducing costly platform efforts to write, refine, and enforce policies on these often highly charged topics.

Second, this move reflects the skepticism many people harbor both towards mainstream media and platforms as fact checkers, a viewpoint highlighted by X CEO Linda Yaccarino in response to the Meta announcement. The role that individual influencers, podcasters, and other nontraditional forms of media played in the recent U.S. election underscores this trend. Users may be more likely to trust context written and validated by their peers, as opposed to an assertion that News Publisher Y or Platform Z deem a given statement to be false.

It’s also usually faster for individual users to write and propose context for a piece of content than it is for a platform, news organization, or fact-checking organization to do the same. Individual users lack the process restrictions in place at such organizations, which increases speed but also significantly increases the amount of opinion-based claims and other noise to be sorted through during the crowd-based note vetting process.

The problem becomes establishing the requisite diverse consensus in order to actually append a note to a piece of content in a timely manner. X has established reputation systems for contributors that consistently rate helpful Community Notes as such, and has made other recent efforts to increase the speed with which helpful Notes are appended to posts. It will be interesting to see how Meta’s approach both aligns with and differs from X’s in this and other regards.

As they roll out, advertisers will need to be mindful of the impacts these changes may have on content monetization and ad adjacency, including understanding (1) whether Meta excludes annotated content from monetization (or makes this content avoidable for advertisers via their platforms’ brand suitability tools), and (2) if so, what threshold of dispute or consensus that content must reach in order to receive said monetization restrictions. For what it’s worth, at Twitter we implemented measures to prevent ad adjacency to Tweets that received Community Notes, and as far as I know, this is still the case today.

Additionally, advertisers will need to clarify how Meta will approach Community Notes on individual advertisements, not just whether they prevent ad adjacency to annotated posts. In 2023, X had several rather public instances of Community Notes on ads, causing at least one impacted advertiser to delete the ad in response. Especially for highly regulated industries such as pharma, crowd-sourced annotations may significantly increase the workload for advertisers monitoring their campaigns on these platforms. Of course, there are clear upsides to allowing users to call out false and misleading claims in ads as well, but the job of advertisers activating on Meta’s platforms would certainly be made more complicated by allowing Community Notes on ads.

Lifting Policy Restrictions

In addition to changes announced to their fact-checking program, Meta has lifted policy restrictions on topic areas including immigration and gender, citing overreach by well-intentioned efforts to be more inclusive in these areas. In the announcement, Meta CEO Mark Zuckerberg refers to complex systems responsible for content moderation, and the significant impact of mistakes made by these systems on free expression — even a single-digit percentage of the time.

Over the years, those of us who have done brand safety work at social platforms have often fielded concerns from the opposite perspective. Many brand safety & suitability solutions exist that tout similar 99+% effectiveness metrics, and the response we often hear from concerned advertisers is that, when it comes to ads appearing adjacent to harmful content, even one impression is too many. When monetization tools are proudly characterized as 99% effective, that may still translate to thousands of unsuitable ad placements for a given campaign. No content moderation or monetization system will ever be perfect, but it’s clear that on both ends of the enforcement debate (those seeking to avoid undue censorship and those seeking to avoid the presence of, or association with, violative or otherwise brand-unsafe content), imperfectly enforced policies can have significant consequences.

As the bar for acceptable speech and conduct on a platform is lowered, the more perfect their brand safety & suitability tools need to be to ensure advertisers are able to successfully navigate the resulting ecosystem in a way that aligns with their values. To their credit, Meta has a strong track record here — their platforms offer both first- and third-party brand suitability tools, and these measures are rated 99% effective by well-respected third parties such as Zefr.

With that being said, brand safety & suitability efforts on platforms have long benefitted from broad-based consumer safety policies. All content on a platform needs to pass the consumer safety check before a determination is made on whether that content is monetizable — the greater the share of content actioned at the consumer safety stage, the lower the volume and fewer the kinds of decisions that need to be made to ensure brand safety at the monetization eligibility stage.

The wider the delta between content permitted on platforms and content eligible for monetization, the harder those brand safety & suitability solutions will need to work, and the greater the scrutiny on those 99%-effective systems and tools becomes.

Changes to Proactive Enforcement

Perhaps the most overused characterization of content moderation efforts used by platforms is that ​“we use a combination of automated and human efforts to enforce our policies.” It’s maddening to be on the receiving end of that line, but it’s true. In most instances, platforms maintain a series of automated classifiers, models, and keyword-based measures to flag potentially violative content. The exact mix varies by policy and by platform. These automated measures often take action on suspected policy violations (giving content either a thumbs-up or a thumbs-down) without a human ever reviewing the content in question. Human efforts are focused on grey-area cases in which the automated system isn’t sure one way or the other. As part of the announced changes, Meta plans to focus their automated efforts on the highest-severity policy areas (citing child safety, terrorism, drugs, fraud, and scams) and raise the confidence threshold for these automated systems to make enforcement decisions autonomously.

For other policy areas, Meta’s platforms will rely on reactive enforcement, requiring users to report content to Meta that they believe may violate these policies. While this change allows the company to focus its proactive efforts (taking action on violative content before users see or report it) on the worst of the worst, it may also increase the burden on users facing harassment or abuse, forcing them to experience and engage with that content more frequently in order for the platform to take action. How these changes will impact user experience on Meta’s platforms remains to be seen, though Zuckerberg acknowledges that the changes will result in the company catching ​“less bad stuff.”

The decisions announced around relaxed content policies and changes to how those policies are enforced may have significant implications for the environment in which advertisers operate, even if their ads aren’t directly next to content they may find objectionable.

A New Regulatory Approach

The E.U. and other jurisdictions around the world have aggressively pushed legislation in recent years aimed at regulating speech and behavior on social media platforms. In some countries, governments have broad powers to order platforms to remove or otherwise censor content they deem inappropriate. In response, many platforms have made country-specific changes to their services or withheld content in individual countries in order to remain available in these areas.

Meta’s announced strategy to partner with the U.S. Government to more aggressively fight these laws may impact user experience on these platforms around the world, either leading to significantly different experiences on any given platform depending on where a user is located, or an inability to access platforms in certain countries entirely.

While the implications extend far beyond advertising, this dynamic may significantly impact the ability of advertisers to reach their customers around the world, and global advertisers will need to track the regulatory landscape even more closely.

A Shifting Onus

Taken as a whole, these changes demonstrate an increasing shift across the online ecosystem away from platform policies and industry norms, and towards individual choice. Users will need to play a more active role in flagging violative content to platforms than they did previously, and will need to actively decide whether they want to view certain kinds of content, such as political content. Platforms like Bluesky even empower users to choose their own algorithms.

On the advertising side, brands will need to (1) develop, maintain, and be prepared to defend their own stances on the kinds of environments and content that align with their values, and (2) implement those stances using the tools available to them, with the kinds of content automatically excluded from monetization on their behalf increasingly limited to the worst of the worst.

Whether this shift towards free expression and individual agency is a net benefit remains to be seen, but it’s clear that people and brands engaging online will need to play a more active role in tailoring their experiences than they’ve had to in the recent past.