Negative news stories about social media platforms appear to be highly effective at pressuring companies into changing their policies.
at the University of Bremen, Germany, and his colleagues analysed policy changes across Facebook, Twitter (now X) and YouTube between 2005 and 2021, and how media coverage from 26 major English-language publications affected their policies. (快猫短视频 was not among the publications picked by the researchers.)
鈥淭here are really significantly more changes in the policies by the platforms in the weeks and months after an increase of negative reporting,鈥� says Katzenbach. For every negative story about a platform, the likelihood of that platform changing its user policies increased by 6 per cent.
Advertisement
To measure negative reporting, the researchers looked at the content of news stories using an AI model that classifies sentiment. The AI had been trained on a dataset of 11,000 news stories about US politics that were annotated by humans.
Facebook was the subject of the most negative reporting, with around 37 per cent of news stories from the 26 major publications being seen as negative. Around 26 per cent of coverage about YouTube and Twitter was deemed negative.
The proportion of negative coverage about all platforms significantly increased after 2018, when allegations about the misuse of Facebook user data by the consulting firm Cambridge Analytica came to light.
鈥淲e鈥檝e seen that platforms do respond to public criticism,鈥� says Katzenbach. For example, in 2014, a campaign dubbed #FreetheNipple pressured Facebook into . However, those rules, and how they are enforced, remain a topic of debate: Meta鈥檚 oversight board in January 2023 following confusion over what did and didn鈥檛 breach policy.
Another example the researchers highlight came during the 2020 US presidential election campaign, when Twitter announced it would ban political advertisements from the platform. Facebook was then criticised in the media for not fact-checking or banning political ads itself, and near the end of the year it too started to do this.
鈥淚 think it鈥檚 very much true that negative media coverage affects the way platforms govern and behave,鈥� says at Northumbria University, UK. Are believes that deciding policies through media coverage alone is not enough to protect users. 鈥淎 strategy that relies on PR is just a Band-Aid approach for the issues that we鈥檙e facing right now,鈥� she says.
Are says a more proactive approach is needed to regulate social media platforms, including legislation that checks how policies affect users. 鈥淚f all we get to do through the use of PR is to change policies, then we鈥檙e taking for granted that platforms are always going to run the system as they are 鈥� while we need to hold them far more accountable than that.鈥�
Katzenbach echoes this view. 鈥淒efinitely you need regulation, and strong regulation,鈥� he says. 鈥淢edia coverage cannot substitute that.鈥�
Political Communication
Article amended on 26 July 2024
We clarified the kind of news stories the AI model was trained on
