Our first analysis of LLM-powered search engines with AI Forensics, in partnership with AlgorithmWatch The evidence collection pipeline has been challenging to build, but we are now ready for more such investigations in 2024! Bing itself is sorry about it 😂

TikTok, who’s there ? This year, in Krakow, I presented the TikTok auditing landscape, from the data access put in place since the DSA, to the adversarial tools that are driving most research. The TikTok representatives in the room « appreciated and will report the feedback » 😄

1/15 For those who believe the Israeli version of the Al Hali hospital attack, I recommend reading this article by Salah Hijazi on L'Orient-Le Jour (very well documented, and much less partisan than Al Jazeera English). Thread to summarize & translate this article

🚨BREAKING: TikTok's 'For You' feed exposes young people to harmful mental health content. Our joint research with Amnesty Tech reveals alarming statistics - after 5-6 hours, nearly 1 in 2 videos are potentially harmful. Read the reports: amnesty.org/en/documents/P…

🤖💡 Plongée au cœur des enjeux de régulation de l'IA. Quelles approches pour encadrer l'Intelligence Artificielle de manière éclairée ? Table ronde sur les futurs contours de cette révolution technologique avec Antoine Bayet Marc Faddoul Nicolas Gaudemet et Thaima Samman. @mediaenseine

Our investigation with AlgorithmWatch into Microsoft Bing Chat revealed that the chatbot spreads factually inaccurate and fabricated information about elections in Switzerland and Germany. Find out more: aiforensics.org/work/bing-chat…

En direct du Sommet Grand Continent. L'impact social et économique de l'IA, avec : • Martin Tisné • Brando Benifei • AnneBouverot • Anu Bradford • Marc Faddoul • Gabriela Ramos x.com/i/broadcasts/1…

Paramount public heath experts perspective in The Lancet on what makes the situation in Gaza genocidal Dramatic amount of red indicators And “Before October's escalation of violence, the effect of the Israeli siege of Gaza had already been described as a slow-motion genocide”

🧵1/6 October 7 was an attack on all Israelis for being Israeli and on Jews for being Jews. It was a pure physical rejection of even legitimate Israeli sovereignty / Dahlia Scheindlin haaretz.com/israel-news/20…

NO EMBARGO IN SIGHT: META LETS PRO-RUSSIA PROPAGANDA ADS FLOOD THE EU Our latest investigation shows a significant loophole in the moderation of political ads on Meta platforms, highlighting systemic failures just as the EU heads into elections. aiforensics.org/work/meta-poli…

Hey Meta 👋 Why did you approve over 95 pro-Russian propaganda ads targeting France and Germany, reaching a total of over 1.5 million accounts in the last week? #noembargoinsight

🚨 After our recent study revealed Meta's flawed moderation of political ads, enabling massive foreign interference weeks ahead of the elections, the EU Commission is launching an investigation into potential Digital Services Act violations.

🚨BREAKING NEWS Following our "No Embargo In Sight" report, the European Commission announces a formal investigation into Meta for potential violations of the Digital Services Act, focusing on political ad management and content moderation. ec.europa.eu/commission/pre…

One+ month since AI Forensics 's damning report on Meta's failure to moderate political ads. Despite the European Commission investigation it triggered, we see no action. Pro-Russia propaganda is in fact growing at an alarmingly rate as the election approach!

« Les parlementaires manquent souvent de compétences numériques. Ils ne sont qu’une poignée à maîtriser le sujet. Il est urgent de combler ce retard. Je veux prendre le lead sur la question numérique à l’Assemblée nationale. » ladepeche.fr/2024/06/11/bap…

Join our online discussion, which is part of the Future of Democracy & Technology Initiative by GLOBSEC, and find out more on how AI is rapidly transforming civil society! Register via: bit.ly/4bURjj7

Fascinating and hilarious thread from Paul Bouchaud , who managed to counter-intelligence the propagandists behind Doppelganger, exploiting a blunder they made!

Curious about algorithmic gossips? Read about our latest research, published and featured this morning in Wired: wired.com/story/tiktok-g…

En même temps.

Are you fucking kidding me this is the best thing I’ve ever seen