BFM TV French Tech: Blue Efficience, a solution to fake news - 03/16
Thierry Chevillard, founder and CEO of Blue Efficience, was Laure Closier’s guest on French Tech this Monday, March 16. On Good Morning Business, he discussed the company’s work specializing in detecting and taking down fake content. Tune in to the show Monday through Friday, or catch up via podcast.
Our French Tech guest this morning is called Blue Efficience. We're with Thierry Chevillard. Good morning. You are the founder and CEO. You work on detecting and taking down fake content. You want to industrialize the process. Honestly there's plenty to do. Especially whenever a conflict breaks out, things explode on social media. You work for broadcasters, content distributors, and TV channels. What's your basic method?
Good morning Laure, good morning Léa. First, thank you for having me. Well, the basic method, first of all, Blue Efficience isn't a new company. It's a company I founded in 2008 that currently offers two solutions. The first one is a copyright protection solution. So on that front, we already work in the audiovisual sector to protect movies, TV shows, and entertainment content. It's all these technologies that also allowed us, in 2024, to build this online reputation protection solution that combines technology and legal tools. The goal is to provide a comprehensive, end-to-end fake content management solution for our clients, from detection all the way to content removal. It's therefore a solution, as you mentioned, targeted at media companies, but also banks, the financial sector, the pharmaceutical industry, as well as institutions, political parties, and individuals. So for you, show hosts and panelists, you're people whose image can be hijacked by cybercriminals to legitimize certain narratives.
But what do you actually do for me? Do you scan social media, scan everything, check where my image might be used? I know there are prime targets,Élise Lucet, for instance. You see tons of promotions using her face for things she actually has nothing to do with, because it's fake. Do you scan social networks?
Exactly. And that brings me to the process itself. It's an end-to-end process that starts with keyword-based web searches. So we search for your names on the internet, cross-referencing them with a wide range of topics. Finance, obviously, given your names and BFM's scope, but also foreign interference, conflict zones, and common scam tactics. All these searches generate a massive volume of results that are then analyzed by our AI. And that's really the core building block we added in 2024: an AI capable of performing up to 1.5 million monthly facial recognitions. It scans these visuals, images and videos found online, to check if you, Laure, or you, Léa, appear in them. From there, it flags items to our analyst team, who identify the infringements and violations present in these visuals to get them taken down. This whole pipeline operates continuously and as fast as possible, meaning some of the content we identify can be taken down within hours.
But aren't you basically setting up AI battles? Meaning one AI is detecting a video created by another AI? Why is that? Is it because humans just can't keep up that you chose to use AI?
AI allows us to handle scale, and that's where industrialization comes in. Very clearly, our reputation protection solution is the industrialization of fake content management for our clients, plain and simple. It's interesting that you mention AI battles, because AI isn't something to throw away, it has plenty of benefits, but it can also be weaponized by cybercriminals. So yes, we need AI to handle the sheer volume of content generated online, and more importantly, how it spreads. Because it's not just about the number of fake contents created, but rather their viral reach, how effectively they gain traction among citizens who end up sharing them.
And how do you manage to get them removed? It multiplies so fast it seems almost impossible.
That's where the legal side comes in, along with our track record as a company operating since 2008. We have extensive experience enforcing regulations. Enforcing regulations. Copyright was simpler, so to speak. When we moved into reputation management, things got more complex. But there are many legal angles to leverage, particularly around trademark protection, right of publicity, defamation, identity theft, and fraud. There are also plenty of legal tools provided largely by the DSA (Digital Services Act), as well as pre-existing laws. We issue notices to platforms and host providers to secure removals, and we achieve a success rate over 90%. Some cases take longer and require stronger enforcement, but the core of our job is to process large volumes as fast as possible through digital notices.
Your technology handles images and video. Does it cover text as well?
Text is very interesting, and we do get client requests for it. It's actually easier to scan text. Right now, we're developing that capability, but our primary focus for reputation protection remains image and video analysis, scanning inside visual media using our proprietary in-house AI models.
What trends are you seeing emerge the most over the last few days?
It's funny, because local elections are coming up and I expected you to ask about that. We have seen fake content affecting our clients involved in municipal politics, but at the same time, there's the war involving Iran. We're seeing foreign interference channels reactivate on Telegram to push fake news reports and narratives aimed at undermining the US and its allies regarding the war in Iran. And there's still that underlying baseline around Ukraine and its allies, claiming they're on the wrong track.
And you have to fight back, respond, get things taken down... Because some companies hit by fake news say, "We don't want to get dragged into a back-and-forth, let's just ignore it, people know it's fake." Are they wrong?
They're wrong. Absolutely wrong. You have to respond and take action, because at the end of the day, it damages the media's reputation. It hurts BFM's brand, it hurts TF1's brand... TF1 and France Télévisions happen to be our clients. It's their brand on the line, and that's what they chose to defend by adopting a reputation protection solution. So taking action to protect your brand image is absolutely vital.