Hate speech, disinformation ahead of South African elections: Big tech makes it harder to police social media

In an age of digital dominance, the movement for access to data is intensifying around the world. Recognizing that data is the new currency of power, advocates are championing transparency, human rights, and electoral fairness. Yet clear imbalances remain. While progress has been made in the EU and US, the Global South remains largely excluded. In South Africa, pioneering research is fighting misinformation hampered by data blockades by tech giants. Growing calls for fairness have given rise to a global movement calling on platforms to democratize access to data.

To stay on top of the content that matters to you, sign up for BizNews Insider’s early morning newsletter. The newsletter will arrive in your inbox at 5:30 a.m. weekdays. Sign Uphere.

by Guy Burger*

There is a growing global movement to give researchers access to the vast amounts of data that are collected and used by digital operators.

Momentum is building because it’s becoming increasingly clear that data is powerful, and access to that data is key for a variety of reasons, including transparency, human rights, and election integrity.

However, there are currently large international asymmetries in access to data.

Some progress has been made in the European Union and the United States: EU researchers studying risks, for example, have a legal right of access, and even in the United States, some companies are taking voluntary steps to improve access.

In the Southern Hemisphere the situation is generally very different.

The value of data access is clearly illustrated by social media monitoring during elections. South Africa is a good example. Powerful “big data” analyticsRecently PublishedIt reported on online attacks against female journalists in the country and warned of an intensification of attacks before and after the May 29 elections.

Ahead of South Africa’s national and provincial elections, many data groups are working to monitor social media for hate speech and disinformation. At a recent workshop bringing together 10 of these groups, participants said they were working to detect coordinated “information manipulation” that could harm the elections, such as through foreign interference.

But researchers can’t get all the data they need because technology companies won’t give them access.

This is something I’ve been concerned about since I first commissioned the Handbook on Harmful Online Content.Journalism, Fake News, and Disinformation: A Handbook for Journalism Education and TrainingSix years ago. Since then, I have overseen a major UN investigation.A balancing act: Countering digital disinformation while respecting freedom of expression.

Over the years, I’ve learned that to thoroughly examine online disinformation, you need to go deep inside the social media engine. Without comprehensive access to the data stored there, we’re left with little insight into the actions of manipulators, the role of misled customers, and the fuel provided by mysterious corporate algorithms.

Monitoring

Focusing on social media in the South African election, researchers at a recent workshop shared how they were doing the best they could with limited data: They were all monitoring text on social platforms, some were monitoring audio, and others were looking at “synthetic content,” such as material created with generative AI.

About half of the 10 efforts tracked followers, impressions and engagement. Nearly all looked at content on Twitter, and at least four monitored Facebook. Three targeted YouTube and two targeted TikTok.

WhatsApp has flown under the radar: Though most messages on the service are encrypted, the company knows (but doesn’t make public) details about who registered users are sending content in bulk, who is forwarding it, whether group admins are active, and a host of other “metadata” information that could help monitors track dangerous trails.

But researchers are unable to do the deep data digging they need. They don’t know how to make sense of the difficult data conditions they are studying.Severely restrictedIn accessing data.

One of the data sources they use is expensive (and restrictive) packages from marketing brokers (who buy data assets wholesale from platforms).

The second source of information is the analysis of public posts online (excluding conversations in groups or on WhatsApp). Using scraped data is limited and laborious, findings are superficial, and risky, as scraping is prohibited by most platforms’ terms of use.

None of the researchers covering South Africa’s elections have direct access to the platforms’ proprietary application program interfaces (APIs). These gateways provide a direct pipeline to the computer servers that host the data. This key resource is used by the companies to profile users, amplify content, target ads, and automate content moderation – a crucial input for monitoring online election damage.

In the EU,Digital Services ActIt allows vetted researchers to lawfully request and obtain free, potentially far-reaching API access to search for “systemic risk” on the platform.

In the United States, things are more open. Meta, the multinational technology giant that owns and operates Facebook, Instagram and WhatsApp,Carefully SelectedIn the 2020 election, 16 researchers (includingJust 5 projectsThe company then outsourced the review of Facebook and Instagram access requests (from anywhere in the world),University of Michigan.

One South African researcher tried the channel but was unsuccessful.

Other platforms such as TikTok still make unilateral decisions about who can access data, even in the US.

Outside the EU and US, it is difficult to even initiate a dialogue with the platform.

Counterattack

Last November, I hosted major technology companies at a workshop in Cape Town on data access and elections in Africa.no response.

The same pattern is evident in efforts made by the South African National Editors Forum earlier this year, which proposed a dialogue on the human rights impact assessment of online risks in South Africa’s elections.Ignored.

Against this background, two South African NGOs, the Legal Resource Centre and the Freedom of Expression Campaign, have launched a comprehensive study of the legal system in South Africa.Information Disclosure Promotion ActForce each platform to disclose their election plans.

However, the two companies refused to respond, citing that they are not subject to South African jurisdiction.The appeal is launchedCompel national information regulators to make disclosures.

Further impetus for change could come from UNESCO.FacilitateInternational guidelines on the governance of digital platforms, which highlight issues of transparency and research access. UNESCO alsoPublishedThe report I researched was titled “Data Sharing to Promote Information as a Public Good.”

Ongoing projectsAfrica Data Access Alliancewhich currently includes five pan-African organisations, and this coalition (of which I am interim chair) is in discussions with the African Union on these issues.

But there is still no guarantee that all this will translate into the platform opening up data to African and global southern researchers.

Read also:

• Star Trek’s utopian influence collides with Big Tech’s reality check: Parmi Olson
• 🔒 Boardroom – Why some big tech stocks are risky in the new AI-driven world
• 🔒 Big tech companies’ AI dominance sparks regulatory scrutiny, as startups turn to big companies for a lifeline

Guy Burger:* Professor Emeritus, Rhodes University

This article was first conversation Republished with permission