While keyword blocking deprives news publishers of ad revenues and advertisers of engaged audiences, brand budgets continue to flow into unpredictable social media environments. Fiona Salmon, Managing Director at Mantis, highlights the double standard that has emerged in the application of brand safety controls.
Brand safety has become one of digital advertising’s default settings. Few campaigns launch without some form of control in place, and the caution is understandable. But the way it is applied across the media landscape remains far from even. On the open web, trusted news publishers are routinely classified as high-risk environments, subject to sweeping exclusions and rigid keyword controls. In contrast, advertisers have shown a far greater tolerance for uncertainty on social platforms, despite the fact that user-generated feeds can shift tone in seconds and offer far less clarity around context and oversight.
This contradiction has quietly reshaped media investment. Credible journalism is often avoided because of perceived risk, while brand budgets flow into social environments where risk is harder to define and even harder to manage.
How a social media crisis reshaped the open web
The roots of this imbalance trace back to 2017, when brands discovered their ads appearing alongside extremist content on YouTube. This caused brand safety to go from a specialist concern to a central industry priority almost overnight; protective measures were introduced at speed, for good reason.
Initially, those controls were aimed squarely at social platforms, where content was primarily user-generated, fast-moving, and unpredictable by design. Over time, however, the same safety frameworks were extended across the open web. Tools created to manage the volatility of social feeds were applied wholesale, even to professionally produced and critically assessed news content.
The result was a shift towards automated keyword blocking as a primary defence mechanism. Terms associated with politics, conflict, health, or social issues were flagged as ‘unsafe’ without nuance being taken into account, leading to a lack of distinction that swept up journalism designed to explain complex issues and scrutinise power. At the same time, social platforms retained advertiser budgets, prized for reach and familiarity, despite persistent challenges around moderation and misinformation.
When keywords replace judgement
This is where brand safety begins to work against its own purpose. Language is nuanced, yet keyword-led approaches strip that nuance away. A journalist’s report on public policy reform can be treated the same as partisan commentary. Coverage of a humanitarian crisis can be blocked alongside sensationalist content. The distinction between risk and relevance is lost.
For publishers, this over-caution suppresses monetisation of legitimate content. For advertisers, it limits access to quality environments that audiences actively seek out and trust. More importantly, it creates a false sense of security. Blocking words does not equate to understanding risk, particularly when ads continue to appear in social feeds where context can change with every scroll.
The imbalance becomes even more pronounced when video enters the picture.
Video, context, and the limits of control
Video is now central to digital media consumption, but it also exposes the weaknesses in how brand safety is applied. On social platforms, video content is often assessed using limited signals, such as metadata, captions, or automated transcripts. But these aspects struggle to capture what is actually happening on screen, particularly when footage is silent, fast-moving, or emotionally charged.
A clip filmed on a phone can move rapidly through feeds, even getting remixed, detached from any explanatory context. The surrounding content can change instantly, yet brand tolerance often remains high because the social platform environment is familiar and scaled. On the open web, video is more likely to sit alongside editorial framing, companion articles, and clear signals of intent. Ironically, and unfortunately, that additional context does not always translate into greater advertiser confidence.
This highlights a wider issue. Traditional brand safety systems are frequently better equipped to scan for surface-level signals than to interpret meaning, sentiment, and suitability. In video especially, where understanding relies on emotion, association, and narrative, blunt controls currently struggle to reflect real risk. The outcome mirrors what happens in text-based news. Quality environments are treated cautiously, while ambiguity elsewhere is accepted as a trade-off for reach.
Towards a more balanced definition of risk
If brand safety is meant to protect brands without distorting the market, it needs to evolve. That means moving away from universal rules and towards a more precise understanding of sentiment, context, and meaning. Advanced AI-powered brand safety solutions are increasingly enabling this, even at video level, allowing precise, suitable ad placements for brands in line with their real values.
The open web depends on advertisers being able to distinguish between content that is genuinely harmful and content that is simply serious. Applying the same safety logic to news and social ignores how fundamentally different those environments are. Journalism often covers difficult subjects, but difficulty alone does not equal danger. Tone, purpose, and presentation matter.
Rebalancing brand safety is not about lowering standards. It is about applying them intelligently. Doing so supports advertiser confidence, protects brand reputation, and ensures that credible journalism is not unfairly penalised for doing its job.
Follow VideoWeek on LinkedIn.
