Public commentary is increasingly being replaced by what has become known as “AI slop”, which Merriam-Webster defined as “digital content of low quality that is produced usually in quantity by means of artificial intelligence” when it made “slop” its 2025 word of the year. By one count, AI-generated articles went from 2.2% in January 2020 to 51.7% by May 2025. Apart from clogging the media environment and taking writing jobs, much of it can’t be verified, and some of it is simply made up. 

This week the Electoral Commission of South Africa (IEC) teamed up with Google, Meta, Microsoft, TikTok and Moxii Africa to fight election misinformation ahead of the 4 November local government elections. The IEC has already received more than 700 complaints of AI-related election misinformation. The platform reported most often is X, which isn’t part of the agreement. “A technically flawless election can still lose public trust if citizens are overwhelmed by disinformation,” said the IEC’s Chief Electoral Officer, Sy Mamabolo. 

The problem with AI-generated online articles, beyond clogging the media environment and taking writing jobs, is that much of it cannot be verified and some of it is simply made up. South Africa’s own draft national AI policy was withdrawn from public consultation after it was found that it contained fictitious references generated by AI. 

Good judgement should be what separates AI-generated content from human reporting, but the sheer volume makes that harder, and the shortcuts we use to cope with this information overload create their own biases. This may look like searching for something online, and the first thing to come up is an AI-generated summary, built from the websites that allow AI crawlers to read them.  

A GIBS study published in August found that only 30.4% of 263 South African news sites block any AI crawler. The summary at the top of your screen is drawing on whoever left the door open, which isn’t always whoever did the best or most honest reporting 

Reading the news well now means checking sources you wouldn’t have had to check before 2022. None of this is new. Before any of these tools existed, reading the news well still meant asking the hard questions that may expose biases. In the interim, we got too used to having that thinking done for us.  

We need to return to the critical thinking skills that can tell the difference between a story of substance and a story of bias. That starts with picking up the old habits of asking who wrote something and what they stand to gain, or noticing when a piece tells us exactly what we already wanted to hear. It also means being comfortable saying we don’t know yet, which is harder than it sounds when every feed rewards having an opinion first. Ahead of an election, this is even more important. The IEC can ask the platforms to take down a fake image, but it can’t make anyone stop and question it before they believe it. That part has always been ours to do, and no partnership or piece of software will do it for us. 

– Erin Dodo
Account Executive