Reducing misinformation and hate speech on social media.

Past and Current Partners

UNDP Sudan, Geneva Centre for Security Policy Prize for Innovation in Global Security, Echoing Green, Unreasonable Institute, Global Social Benefit Institute (GSBI), Accelerate2030 Fellowship, GSMA Connected Women Award, Stevie Award for Women in Business, SPRING Accelerator

Active Countries
Sudan, Sri Lanka, Myanmar
Thematic area(s)
Crisis, Other
Organisation Name
Koe Koe Tech Foundation

The Problem

Social media platforms need to flatten the curve of dangerous misinformation. Hate speech, incitement of violence, and fake news are a problem for 190+ countries in 200+ languages. Content moderation in the global south is inefficient and ineffective as CSOs do the heavy lifting on content moderation meaning that individuals must sift through millions of posts daily. CSOs unfortunately have no standard way to do this, which makes reporting inconsistent and little feedback from social media companies whether or not posts are ever removed.

The Solution

A webplatform that uses AI/ML to flag hate speech, violence, misinformation, and disinformation in real-time from posts and comments from Facebook, TikTok, and Twitter. The solution offers end-to-end content moderation, fact-checking, data dashboards, IHRL-algorithm prescreening, and training dataset generation, powered by local CSOs in local languages. CSOs can work more efficientially and social media can remove content with more confidence. Reports are sent to social media companies, which report back which harmful posts are actually removed. In a pilot in Sri Lanka, 91% of reported harmful posts veriffied have been taken down from TikTok and 58% taken down from Facebook.

How it works?

  • Step 1: Approaching local offices of UN agencies and well-known local CSOs that are specialized in fact-checking and fighting hate speech to demo the software.
  • Step 2: Once an MoU is signed and the funding is confirmed, onboarding begins and accounts are created for the local UN offices and/or the local CSOs.
  • Step 3: CSOs and UN offices provide a list of Facebook pages and Twitter accounts that they consider generators of hate speech and misinformation and a dataset of the local language to start training the software platform in the new local language.
  • Step 4: Once the user account is set up, they are provided with training on how to use the platform.
  • Step 5: The local CSOs and UN offices start using the platform, which allows them to automate their processes for reporting hate speech and misinformation to the social media platform and generate policy reports to advise local and international policymakers.
  • Step 6: Monthly meetings to collect user feedback on the platform and its features as well as detect possible bugs.
Digital X Solution Meekin

Bridging the digital divide

Since 2019, they have been delivering digital literacy, online safety, and misinformation training to smaller CSOs and to young people in Myanmar with a focus on those living in IDP camps.

Impact and highlights

During an election in Myanmar, 7 CSOs verified 4,000 harmful Facebook posts or comments in local languages, flagged by Meekin's AI/ML (of which 50% were taken down by Facebook). In a pilot in Sri Lanka, 91% of reported harmful posts veriffied have been taken down from TikTok and 58% taken down from Facebook.

Plans for expansion

Seeking to implement the platform in Turkey and Morocco, where they have strong relationships with potential local partners.