The next UK general election will be held on July 4. There are plenty of ways to get informed on who to vote for, including browsing manifestos, taking in debates and reading trusted sources.
It’s easy to become derailed by misinformation, though. And, in nastier campaigns like this one, you may even hear misinformation parroted by officials and other public figures.
Earlier this year, a survey of IT workers by BCS, The Chartered Institute for IT, found that more than half of professionals in the field feared AI deepfakes could affect the results of the upcoming general election.
Here are some tips to help stop such questionable content from making its way from your social feeds, and other news sources, and into your head. We’ll also look at some resources that can help in uncovering the truth of events misinformation seeks to mine for engagement.
Don’t trust social media posts or ads
Spreaders of disinformation on social media use sneaky tactics that exploit the way most of us scroll through feeds – rapidly and without all that much care. Sometimes, nefarious posters will claim to speak for trusted figures, Martin Lewis being a key example used in the past, or ape the style of respected sources like the BBC News website.
Social media platforms’ control over the quality and veracity of ads and sponsored posts is poor, meaning all sorts of stuff end up displayed.
The simplest solution, if you want to find out if a publication or person actually did make a claim, is to go to the source. Clicking through from these social media posts or ads can often take you to a spoof website, so avoid doing so.
Be wary of AI-generated audio and video
One of the big fears for elections across the world is voters will be flooded by deceptive content made by AI. It can fake real figures in audio, and video, often with alarmingly lifelike results. These are known as deepfakes.
We’ve seen less of this than feared so far, at least in the UK. And, as there’s no magic digital black light you can apply to this stuff to reliably identify AI-generated content, it’s yet again important to be extra-mindful of where this content has arisen. Is it a reliable source or simply a repost from someone online?
MIT does offer a series of questions to ask yourself when you spot anything suspicious. The deepfakes of today will often not have the lighting or glare you’d expect of a real interview.
“Pay attention to the eyes and eyebrows. Do shadows appear in places that you would expect? DeepFakes may fail to fully represent the natural physics of a scene,” it says.
However, as AI video generation improves, these tips may fast become out of date.
Editing, not AI, may still be the biggest problem
This issue of potentially misleading editing has cropped up more in the US pre-election coverage than the UK’s of late. It’s always important to consider the context of any clips you see or hear online.
The narratives of reality TV shows are constructed using judicious clipping, messing with the chronology of content and using music to massage your reaction to it. Cropped and framed videos and audio can be used to skew someone’s interpretation of an event, too.
For example, during the G7 summit, clips appeared to show Joe Biden aimlessly wandering off from other world leaders during a parachute stunt. However, out of shot, there were actually other parachute jumpers Biden was addressing.
Use fact-checker resources
There are several fact-checking organisations in the UK, and all of them are putting significant efforts into the UK elections.
Full Fact is the best-known. It’s a registered charity funded by donations, and has been around since 2010. The Full Fact blog digs into claims made by the parties.
BBC Verify is a strand of the BBC’s news coverage launched in 2023. It’s a team of around 60 journalists that does “everything from open source journalism and verification to data journalism, disinformation, analysis and fact checking”.
Channel 4’s fact-checking initiative was among the very first, at almost 20 years old. PA and Reuters also have fact-checking sections.
If in doubt, head to one of the resources to see if they have covered a topic.
Are these resources flawless? With limited resources, there’s going to be at least potential for bias in what they choose to cover. But they can be considered a gold standard in terms of online veracity at present.
Community Notes can be useful
X, previously known as Twitter, also uses a system called Community Notes to help combat misinformation and potentially harmful content.
People sign up to become a Community Notes contributor, and folks from the UK became eligible in January 2023. These notes appear below contentious posts, and have been useful in identifying when, for example, footage from an older event purports to be from a breaking news story.
YouTube has also announced plans to introduce a similar viewers' notes system, although it’s unlikely to hit the UK before the next general election. Roll-out starts in the US.
It’s for “notes that clarify when a song is meant to be a parody, point out when a new version of a product being reviewed is available, or let viewers know when older footage is mistakenly portrayed as a current event,” according to Google’s blog.
Don’t rely on chatbots for information
Whatever you do, don’t rely on AI chatbots to get up-to-date information. These are subject to what are known as hallucinations, where the AI makes things up.
For example, Sky News found it could get ChatGPT to say Labour had won the election – complete with stats – a month before the actual election was to be held. These hallucinations can be caused by a chatbot picking up on fake or misleading content posted online, or by creatively filling in the gaps where information wasn’t available at all.
“If the training data is incomplete or biased, the AI model may learn incorrect patterns,” says Google. A chatbot can at times be about as reliable as a know-it-all friend spouting off in a pub after a few drinks.
