GARM’s Shutdown Is an Opportunity to Finally Get Brand Safety Right

It’s been a strange few days for brand safety, from the initial news of Elon Musk’s X lawsuit against the Global Alliance for Responsible Media to the shock that GARM would be shutting down.

What’s surprising is not the fact that it had happened—as a podcast advertising leader, I’ve seen firsthand how the org’s approach to brand safety has served neither advertisers, publishers, creators, nor the medium itself. However, as much as the ad industry needs to find a new path forward, I never expected the collapse to happen so suddenly, or for the reasons stated in the press.

After years of being the default authority for brand safety online, GARM had recently been criticized by right-wing groups for supposedly silencing conservative voices. While I’ve been vocal that GARM is an imperfect mechanism for promoting brand safety, the arguments that conservative groups are making don’t add up.

The GARM framework failed consistently in application, with brand safety tools like Barometer and Sounder producing outputs that were destructive for content on all sides of the political spectrum. This led to shows on the left and right being inappropriately graded, costing both sides vital advertising dollars.

The sad truth is that GARM’s shutdown is the right thing, but possibly not for the right reasons. However, it does present an opportunity to build a new approach to brand safety. First, though, we need to truly understand the nature of GARM’s shortcomings, divorced from the political spectacle surrounding it.

Why everyone loses

Despite the protestations of some conservatives, I trust the intentions of the World Federation of Advertisers (WFA) and the individuals who worked on GARM.

It’s not that its standards are somehow biased, but they are so impractical that they hurt everyone—right, left, and center. Too often, GARM asked the wrong questions, monitoring highly nuanced concepts that are easily misunderstood even with advancements in AI.

Then, their unreliable data is treated as fact without anyone looking under the hood to see if the outputs work. For example, if you use any of the major brand safety measurement tools and compare a podcast’s rating to the transcripts that informed that rating, you’ll see how the GARM system was error-prone.

At the heart of GARM’s brand safety failures is the rise of negative partisanship. The polarization of online discourse is a broad issue, with cultural drivers that are hard to define via contextless keyword automation. GARM’s attempts to use this method have only complicated the issue.

GARM’s framework penalizes content that tackles subjects that would normally be non-issues for most ad placement, including terrorism, spam, and online piracy. And when they address polarization more definitively by tracking relevant metrics like “debated sensitive social issues,” the available market solutions still rely on keyword detection for potentially controversial topics (e.g. Biden, Israel), slapping highly relevant and responsibly produced news items with labels like Medium Risk or High Risk, regardless of how carefully those topics were discussed.

GARM indeed has some roots in politics as a byproduct of Donald Trump’s political success, as well as subsequent claims of misinformation and disinformation. But now, those same accusations are levied against both liberals and conservatives, muddying the waters. GARM tried to address misinformation in their framework but then realized it is not something that any person, organization or tech can do in a way that would be acceptable to a broad group of people.

Ultimately, for marketers who don’t want to get caught in the middle of the culture wars, the choice becomes simple. It’s easier to not advertise on any show or platform that could be unsafe, rather than to risk being the next Harry’s Razors, accused of supporting harmful ideas or political favoritism.

But this hurts the entire market, regardless of political affiliation.

Finding civility

GARM’s thesis began as a positive step in applying nutrition labels to media, a standard the industry still desperately requires. It just needs a different approach.

No one cares about differing opinions when it comes to brand safety. But when personal attacks get involved, brands get caught in the middle. This is where future brand safety efforts should focus—trading measurement of contextless keywords for measurement of these types of ad hominem attacks that degrade civility.

If we can measure personal attacks and make brands aware of the civility levels in environments where their ads may appear, we can rise above accusations of political bias and promote a healthier online discourse while creating a safer, more reliable way for brands to share their messages.