Twitter is the most brand-safe platform for advertisers to appear on, according to the findings of a new brand safety study carried out by computer vision firm, GumGum.

The company questioned 274 executives from brands, agencies, publishers, and technology providers in November 2018. 45 per cent of respondents called out Twitter as the most brand-safe platform, with Facebook second on 25 per cent and Instagram third on 17 per cent. Snapchat and YouTube were joint bottom of the list, with just 0.38 per cent of respondents naming them as the most brand-safe platform.

A majority of respondents felt that brand safety had improved over the previous 12 months, with 49 per cent saying the problem had got “much better” and 32 per cent saying it was “better”. The main area of improvement was less exposure to brand-unsafe images, cited by 80 per cent of respondents. 57 per cent of respondents said they had created an internal role or roles specifically dedicated to preventing brand safety incidents in the past 12 months.

When asked about the tools deployed to mitigate the problem, the most popular answer was blacklists (62 per cent), followed by natural language context detection (35 per cent), and image recognition (31per cent).

Violence is considered the most brand-unsafe type of content to be seen adjacent to, cited by 37 per cent of respondents. Amusingly, second in the list is competitor’s branding on 22 per cent. This is seen as much worse than appearing alongside pornography (3 per cent), fake news (2 per cent), or bad news (0.38 per cent).