Investigative journalists in Spain proved that TikTok’s own safety system does not work. Google needed 2 years of constant reporting to act on their safety regulations. And in Brazil big tech lobbying actually influences legislation through buying trips to the Caribbean for congressmen. As digital platforms claim to protect users from harm, investigative journalists increasingly test those claims by auditing moderation systems, reporting workflows, and recommender behavior.
Can journalism hold big tech accountable? A recap of the discussion “Testing the platforms: investigative journalism, accountability and policy impact” at the International Journalism Festival in Perugia.
CAN INVESTIGATIVE JOURNALISM HELP IN A FIGHT WITH ‘AI SUPREMACY’?
Investigative journalism hand in hand with advocacy generates a new way of creating impact for newsrooms, a moderator of the panel, director of Information Integrity Initiative Julie Posetti is sure of. Julie (PhD) is a multi award-winning internationally published Australian-British journalist and academic. Information Integrity Initiative is a project of TheNerve – the digital forensics lab established by the Nobel Peace Prize winning journalist Maria Ressa.
‘Thanks to investigative reporting and whistleblower revelations we now have ample evidence to illustrate the impunity with which big tech companies act with regard to the safety of their users and the corrosive impact of their systems on democratic societies and institutions. But even as they roll back their Trust&Safety teams and acquiesce to the authoritarian demands to eliminate systems designed to minimize disinformation, hate speech and abuse on their platforms. They gaslight us by claiming that they’re still working hard to protect users from harms inflicted by their technologies, while profit is clearly an overriding objective and some say the only priority’, Julie says.
It’s imperative that investigative journalists increasingly test the claims of these platforms by auditing moderation systems, reporting workflows and recommending behaviour. But that’s not enough without also tracking political influences, Julia Posetti notes.
‘Investigative journalism working cooperatively with advocacy systems can generate a new way of creating impact for newsrooms that goes well beyond audience reach. Controlled testing, transparency of methods and human safety framing can translate journalism into accountability work without sacrificing editorial independence’, she claims.
So let’s dig into three case studies from Europe, US and Latin America that show us how to make platform investigations credible and to feed our findings into advocacy and regulatory efforts.
USA: HOW TO FIGHT ELON MUSK AND GOOGLE
‘We’re using the word disinformation a lot. Word that’s essentially been banned by US political forces with reference specifically to foreign policy. How do you continue to try and achieve accountability for the facilitation of disinformation on these platforms’, Julie asks one of the panelistis – Alexios Mantzarlis. He’s a director of Security Trust and Safety Initiative Cornell Tech and a co-founder of the tech publication Indicator. Prior to this Alexios Mantzarlis worked 5 years at Google Trust&Safety, trying to improve digital safety across the online ecosystem.
What does it actually take to affect policy change within a big tech company? How much harder has it become in recent years?
‘Trust&Safety teams across industry are composed of people who care. Over the past 2 years we have seen a concerted political effort to dismantle those teams. It is no coincidence that the Elon Musk Overton Window (developed by Joseph P. Overton, this model explains that political viability shifts; ideas can move from unthinkable to mainstream) shifting phenomenon of making ‘might is right’ (‘Might Is Right or The Survival of the Fittest’ is a book by Ragnar Redbeard, that argues that only only strength or physical might can establish moral right) on platform started with going after the Head of Trust & Safety. The next Head of Trust&Safety and the next. Until there was pretty no one left’, argues Alexios Mantzarlis.
A month after Elon Musk took over Twitter, in November 2022, Twitter’s Trust&Safety head Yoel Roth left. Then in June 2023 second head under Elon Musk Ella Irwin resigned. It was right after the social media platform pulled out of the European Union’s voluntary code to fight disinformation.
‘Behind Musk’s bluster a lot of the other platforms were happy to shed that. Especially as they barrelled towards AI supremacy of the market’, disinformation expert boldly says.
He thinks that despite the situation being pretty grim, there is an extraordinary role for investigative journalists to have an impact. ‘I left Google in part because of a tepid response to what was evident to be the biggest harm of deep fake yet, which is non-consensual intimate imagery’, he shares.
The number one harm, Alexios Mantzarlis claims, is what’s happening to young women and girls around the world with AI nudes. In the past 2 years writing about it at Indicator they had tens of thousands of ads removed, as well as apps. But internally at Google it took 2 years to persuade top execs to act on their own rules. ‘I think the very top of the platform leadership is rotten. I don’t think it’s true across the board. What journalism can do is help with the triaging on getting to the decision makers’, he assures.
Julie Posetti adds that political will to address these problems increases due to the investigative reporting on the issue. Reporting triggers political action that puts pressure on the platforms. Alexios Mantzarlis advises to anchor reports on explicit policies tech platforms already have and potentially their regulatory obligations. Journalists can also equip regulatory bodies with evidence.
‘During the Biden administration the rule was simple: do whatever you want with AI. Just put some labels on it. But even that they’re failing at. We have conducted multiple audits on Indicator and found the very low performance of this rule. But because this is part of AI act regulation, part of Californian requirements producing this body of knowledge helps the decision makers actually prod and push and confirm as the law gets implemented that it’s happening’, he notes.
There should be other auditing mechanisms, Alexios says, but for now all we have is investigative journalism. It also frames the impact of your reporting far beyond audience reach, Julie adds.
SPAIN AND TIKTOK PREDATORS
Co-founder and CEO of Maldita Clara Jimenez Cruz steers a discussion further. What if the platform’s system is flawed? Is there anything we can do if the company doesn’t act on their own rules and policies.
Maldita is a Spanish foundation and nonprofit news organization created to fight disinformation and lies in public discourse through fact-checking and data journalism, which won the innovation European Press Prize in 2021 for its WhatsApp chatbot. Clara’s team published TikTok predators’ investigations on sexualized images of minors. It had a huge impact in Spain with the government instructing the prosecutor’s office to sue the platform.
Clara Jimenez Cruz proposes two approaches for the investigative journalists depending on legal obligations in your country.
‘You can either go to TikTok and give them all the accounts that are promoting AI generated videos of sexualized minors and the sexualized videos of real minors that are on those accounts as well. So that they take them down. Or… We chose a second way. We wanted to test whether the systems that TikTok has in place, when reporting as a regular user, worked or not. For me, it’s not a solution that we have to flag something to a platform for it to be taken down. The solution is to make the system secure enough’, Clara argues.
They reported the accounts that posted sexualized images of minors through the channel that any user can use. In EVERY reported case in exactly 30 minutes the team received an answer saying that these accounts and this content does not breach TikTok policy guidelines.
‘Which is not true. You’re not allowed to have AI generated sexualized images of minors on the platform. So the system didn’t work’, Clara Jimenez Cruz concludes. So then the team reached out to TikTok, told them they were gonna publish this investigation. Only then TikTok asked for accounts to check. ‘We were like: you have the accounts, we reported them using your system’, she proceeds with the story.
After the investigation was published TikTok called Maldita to an emergency meeting on a Friday afternoon with the Head of Trust&Safety Global. ‘We told them once more: you have the accounts. It’s been reported with one user. 15 accounts. We gave them all the dates. To this day they haven’t found the accounts. It’s been 4 months’. Clara’s team had several meetings with TikTok afterwards to discuss that the system definitely is not working. Tech execs assured they are looking for the solution. Ironically, the accounts Maldita informed about, predatory accounts targeting children, are still up and running.
Julie Posetti suggests: the responses you get while reporting policy breaches on TikTok are probably automated. In their report Maldita journalists even gave instructions on how to look for such accounts. For example, searching for keywords like ‘schoolgirl’. But still – nothing is done.
But the investigation had another outcome – legal one. Maldita is a non-profit, that has a newsroom. They divide the advocacy and the reporting teams, who work in parallel during investigations.
‘The way we are gathering the evidence for the reporting can also serve the advocacy team afterwards. And we end up being evidence providers for the law enforcement. Not only to the Spanish government, but also to the European Commission in applying the Digital Services Act’, Clara explains. Digital Services Act is an EU regulation that entered into force in 2022, establishing a comprehensive legal framework for digital services accountability, content moderation, and platform transparency across the European Union.
‘These kinds of things are, at the end, systemic risks. It’s not just a guy sexualizing minors on TikTok. It’s a system that enables it to happen over and over again. The prosecutor in Spain is looking into it. I think the European Commission is gonna be looking into it as well’, Clara Jimenez Cruz optimistically says.
Starting strategic litigation can be one of the ways investigative journalism makes an impact. ‘Even if you do not have a Digital Services Act, you’re not in the European Union, strategic litigation is what makes these things move on. Even if there is no political will’, she concludes.
BRAZIL AND BIG TECH LOBBY
One of such countries that do not have their own regulations is Brazil. Moreover, in Brazil big tech lobby is shaping lawmakers’ positions on AI and social media platforms regulations.
Co-founder and executive director of Aos Fatos (an award-winning organization based in Brazil focused on AI and tech-driven journalism to combat disinformation) Tai Nalon explains that since 2020 there have been different attempts to regulate big tech companies in Brazil.
‘The Fake News Bill’ was problematic at the beginning. But it was adapted after discussions at the Senate and the Chamber of Deputies. And was supposed to be systemic regulation of the market and the content. But it was hindered due to the lobby of big tech companies throughout 2022-2023’, Tai Nalon explains.
On 8 January 2023, following the defeat of then-president Jair Bolsonaro in the 2022 Brazilian general election and the inauguration of his successor Luiz Inácio Lula da Silva, a mob of Bolsonaro’s supporters attacked Brazil’s federal government buildings in the capital, Brasília. This sparked a new discussion about the need to regulate disinformation.
‘But it didn’t work. Instead we have environmental regulations altered for big tech companies to install their data centers. For example, they promise false advantages to local communities. In our investigation we found out that legislators have been traveling by invitations of big tech companies. There were at least 83 visits of big tech representatives to members of the Congress in a year. And those members were responsible for 15 different alterations to the current bills being analyzed in Congress’, says Tai Nalon.
Big tech has to partner with local companies to work in Brazil. So these local companies also become intermediaries between big tech and Brazilian politicians.
‘And Latin America overall. It’s not just a Brazilian problem. Unfortunately there is no framework in Brazil for lobbying. It can be everything and nothing at the same time. One day you travel to the Caribbean having all your wishes paid for. In 2 days you give a speech against the disinformation bill. Is it lobbying or an influence campaign?’, she poses a question.
This case shows us that investigative journalists can also look for political agenda while researching the anti-discrimination legislation. Showcasing the dark side, the money and resources journalists can make the public aware what actually influences their safety online.
MAKE DISINFORMATION REAL AGAIN
Julie Posetti summarizes the discussion with an important notion. We now live in an era of distrust in the overall necessity to fight disinformation. It started in the US for political reasons, and became a global phenomenon. ‘Almost an allergy now to the word itself’, Julie Posetti jokes sadly. At the same time global tech is only growing its influence.
She asks: how can investigative journalists and researchers of the very well documented and scientifically valid field of disinformation studies make an impact regenerating the will to act?
‘It’s not lost to the irony that the regime that was all about free speech has a very clear list of censored words that cannot be funded by the NSF. That cannot be used by the researchers without repercussions. That leads to people losing their visas and their jobs’, says Alexios Mantzarlis. He is convinced: we need to stick to the words that mean something. ‘I’m going to continue using the words ‘disinformation’ and ‘factchecking’. 10 years ago we saw how popular factcheckers were, then – rapid decline. This tells us: it’s a concerted attack. Just like with ‘climate change’. You cannot win a battle conceiving what you’re doing. I think you need to step up and be proud’.
Through three different case studies from different parts of the world presented at the International Journalism Festival in Perugia, Italy we have seen that investigative journalism nowadays is one, if not only, tool in safeguarding the public in the disinformation era. To sum up:
- Investigative reporting increases the political will to address problems of deep fakes propaganda or sexualized AI generated content. Reporting triggers political action that puts pressure on the platforms.
- Journalists need to anchor reports on explicit policies tech platforms already have and also equip regulatory bodies with evidence.
- Starting strategic litigation can be one of the ways investigative journalism makes an impact.
- Showcasing inside political agenda helps to make the public aware what actually influences their safety online.