A new approach to tackling disinformation

Online disinformation is spreading rapidly across social media and digital platforms. Facebook and Twitter are inundated with inaccurate stories about coronavirus (COVID-19). We are all exposed to political bias and conspiracies on a daily basis including inaccurate narratives around 5G and vaccines.

A global problem

Survey data from the 2019 Indian Election suggests that 53% of the electorate were exposed to political disinformation, and email hoaxes have led to civil unrest in the Ukraine.

Although Facebook have taken the decision to ban misleading coronavirus ‘cure’ ads, and Twitter are experimenting with a colour coding system to flag tweets disinformation remains an active threat to democracy, public health and economic prosperity.

The increasing sophistication of deepfakes, shallowfakes, generative imagery and synthetic audio are also coalescing to undermine our perceptions of reality and what can be trusted. Technology is rapidly outpacing regulation, with AI created image profiles achieving Twitter verification.

The Avatarify project even allows for live deepfakes to appear in Zoom video meetings.

How do we fix this?

Social media companies typically see increased use of AI and human moderation as the solution to the problems. Although this approach may rectify some of the issues, the underlying motivations for disseminating inaccurate information (political, economic and criminal) will not disappear.

Ultimately the responsibility is on all of us as social media users to prevent our channels becoming fake news ghettoes populated by extremists.

Introducing CREPT

I am proposing a new model to help assess active risks, and provide an easily understandable overview of content and help us all make informed decisions.

This is based on five factors:

• Credibility What is the source of the story? Can it be verified and trusted? A fully sourced article published by the New York Times would carry more weight than a blog post without attribution.
• Reach How widely is the news circulating? Is it a local phenomenon or global?
• Engagement How are people interacting with the story? Is it being actively shared or commented upon?
• Propaganda Does the story advance a specific political agenda? Is it being coordinated by a national government or unspecified financial backers?
• Threat level Are there significant risks to public safety, health or wider society resulting from the story? Has the story been published with the intention of causing harm or inciting criminal activity?

Key principles of CREPT

• CREPT is an open system that invites scrutiny and review
• CREPT Ratings are peer reviewed to prevent bias and potential infiltration by bad actors
• Evidence based
• Platform agnostic
• Apolitical
• Not affiliated to or funded by any specific organisation

Please visit the CREPT pilot pages to learn more about the project.