Exclusive:
A technology expert has told how easy it is to make fake videos with avatars that spread misinformation and warned that many members of the public are unaware of it
People are being duped by fake footage on social media and unaware how easy it is to make, says an expert warning of the dangers of artificial intelligence.
Making an avatar of yourself or anybody is actually very easy to do and can be done for as little as £30, says Dr Fabio Goncalves de Oliveira, of Henley Business School.
He says that many people do not know about the technology and so are easily fooled by others who are actually quite smart at making images and videos that can be used for malicious intent.
At the same time while he admits that TikTok may be the worst for spreading fake news, he sympathises with social media companies as they exist to spread news quickly and so it is very difficult to stop bad actors. And TikTok “is the worst” because it is also the quickest, he told The Mirror.
Social media has been in the spotlight recently with rioting following the Southport stabbings which is believed to have been stirred up on platforms including TikTok and X.
Disinformation, or deliberately false details including claims that the suspect in the knife attack was an asylum seeker has been spread while AI has also been used to depict Muslims waving knives with these posts shared thousands of times.
Dr Goncalves de Oliveira, who lectures on entrepreneurship and innovation, said how easy it is to make fake videos. “I have an avatar with my image for just £30 which can use my voice,” he said. “It is easily available for just £30 where you can have an avatar and make a video, you just need to download it and it is being used by people around the world.”
It is easy to see the potential dangers from the software. “In schools youngsters are making nude images with the AI and uploading it on Tiktok,” said Dr Goncalves de Oliveira. “Smart people are using it to influence people but it is not widely known which is dangerous. I can tell you I have 100 students and if I ask them, ‘do you know about making fake videos?’ And maybe two or three know and these are university students.
“You take bits of videos and speeches, take out what is said and put it in a different context. We need upskilling so people are aware of what is happening so they are not so easily taken in.”
He said how he agrees with a 2022 academic paper that pinpointed the three areas “where actors take advantage of to propagate fake news” – disasters, health and politics. “These are the areas I think where social media is used and in the case of Southport it was political,” he said, while the way to combat it is through financing and human resourcing.
“But this kind of news on social media travels very fast so it is difficult to stop. Social media uses algorithms to spread the posts faster, that is how they work and that is what people want. Platforms are designed so that people get posts like those they have seen before. So if you follow football all the posts are related to football. People get the information that they want to receive.
“During Covid the WHO said we are not just fighting a pandemic but an infodemic with lots of fake information being spread. There was a 13 year old who said he wouldn’t wear a face mask because he’d seen an influencer who said they don’t work.” Going into the detail of how to stop the fake news, he said that platforms are often playing catch up to suppress information. “You have content moderation and enforcement. So for example Facebook has algorithms to identify bots, flag content that is harmful or fake and then human moderators to fact check,” said Dr Goncalves de Oliveira.
“One person sees a post and flags it as harmful so they stop the spread. Algorithms label fake content by marking it as ‘relevant’, and this is a similar process across platforms. The problem is that trendy topics spread quicker. Enforcement is through removal, warnings, reducing visibility which can be done both automated and through humans. It is very difficult for social media companies to control misinformation with ambiguity and confirmation of bias.
“People are naturally more likely to read things that reaffirm their own beliefs and reject the opposite. It is difficult to control as people head to that information. Different social media platforms are technically very similar, they label stuff as fake, there are human moderators and they delete content.”
As well as having algorithms which act faster to spread news, TikTok also has a demographic of younger people who are more easily influenced, said the expert. “TikTok is worse because of the demographic, 67% of users are 18 to 24 globally and this is the most vulnerable target group as they absorb misleading content,” he said. “They are also the most exposed as they use it more, they were born with a phone in their hand. Social media is their form of communication.
“Tiktok algorithms are faster for propagation and broadcasting, displaying to other posts is through algorithms and the network effect faster. It is faster than Facebook, Instagram and even Youtube. Tiktok can reach a huge audience in milliseconds. So moderation is slower than propagation they can’t keep up. You need to go after to make corrections.”
It is the speed of getting to the malicious actors that is the problem as the location can be identified. “We are able to identify malicious users through IP addresses and ID of mobile phones, networks, we know everything and so it is about speed to go after them,” said Dr Goncalves de Oliveira. “If it is from abroad then we know if it is from a certain country that misinformation is being propagated but we don’t block a country’s content.”
Meanwhile, Telegram boss Pavel Durov has been arrested and investigated for running an online platform that allows illegal transactions, child sex abuse images, drug trafficking and fraud. Claims that Durov’s lawyer said were “absurd”. Dr Goncalves de Oliveira said that Telegram has been used by conspiracy groups and extremists but the messaging app is no different to Whatsapp in terms of its encryption. He said: “There is no difference in terms of encryption with say Whatsapp but you get bigger groups with Telegram, maybe millions – maybe 10million. Foreign actors, for example, will target certain groups through algorithms, they will look for certain information, and feeds a channel with misinformation to fool people.”