For the last 12 months, we’ve seen multiple ad platforms suffer in the limelight because of a breakdown in brand safety. The issue has become more prominent due to the rise in user-generated content which can be posted any time, with the publisher having little control. YouTube is at the forefront of these brand safety scandals and has since stated that AI (artificial intelligence) has allowed them to remove unwanted content quicker than a human would, resulting in more than 150,000 videos being removed from the site in the last 6 months. This is just one example of how AI has been very beneficial in terms of brand safety.
Brand safety is, in some sense, a subjective area where one size doesn’t fit all. Whilst most brands will want to avoid the likes of extremist content, there will also be more brand specific areas that need to be avoided. Current solutions mainly include negative keyword lists, whitelisting and blacklisting. The issue is that whilst this provides some form of safety net, it does have holes. Some ads do fall through these holes, whereas others are caught in the net that is perfectly safe. AI could provide a way to fill these holes by tapping into semantic analysis and language processing to better understand the context of a page, and as a result, help brands breathe more easily.
All Response Media viewpoint
As an agency, we take the safety of our brands very seriously. AI technology at a glance can sometimes be scary and even now, we still have those ‘human’ checks in place when dealing with AI. There’s a perception that increased automation can be dangerous, however, we should realise that AI technology is in our control.
So, can AI really tackle the big issue of ad fraud? In theory, the answer is ‘yes’, AI has the capability to let brands breathe and not worry about appearing against unsafe content. Semantic analysis can provide full context surrounding an ad placement, reducing the risk of appearing against unwanted content and even helping brands not miss out on an opportunity to reach the ideal user with the ideal ad placement. However, what we need to consider is that fact that all brands are different, and to make it work for our brands, we must understand that individual brand and what ‘brand safety’ means for them.
In the meantime, and whilst AI is developing more and more, we currently run our digital activity through MOAT, a brand safety and fraud detection partner. On top of brand safety and fraud detection, this partner also offers real-time display and video analytics as well as direct integrations with social channels.