Sizmek has launched a service within its Peer39 machine learning and semantic advertising offering that allows brands to block their ads from appearing against fake news and satire content.
The ad tech outfit announced plans for the product at the end of 2016 but is now opening it up to brands, via its DSP partners.
A large part of the development included identifying and monitoring websites in order to categorise them under the ‘fake news’ or ‘satire’ label. According to Sizmek, to compile these websites they had to draw from other resources, such as the efforts made by anti-media bias institutions, apps and browser extensions, and compilations gathered via crowdsourcing.
The lists will be updated monthly, which Sizmek believes is sufficient enough as a frequency because the process of monitoring the lists is manual and labour intensive.
As part of the development, Sizmek had to categorise both types of content and the differences between fake news and satire. It said: “The difference is the intent behind their stories. Satirical news sites are explicit in their intentions to humorously mock current events. They often have a disclaimer alerting readers of the satirical bent to their news items. On the other hand, Fake and Partisan News sites do not go to such lengths. They typically present as legitimate news organisations and can only be exposed as fake after outside fact-checking.”
The tool allows brands to opt out of fake news placements at a pre-bid level, as part of its Peer39 offering. DSPs that will be allowing this include AppNexus, Turn, TubeMogul (Adobe Advertising suite), Yahoo! and The Trade Desk.