We are proud to announce the release of our (ACL 2026 Oral) benchmark, VeriTaS, for measuring the fact-checking capabilities of LLM agents! In a world of growing fake news, we believe that the ability to verify claims is critical. VeriTaS is multimodal, multilingual, and made of real-world claims.

Attention #NLProc researchers, the EACL 2027 website is officially LIVE: 2027.eacl.org! 🎉 🇬🇷 Join us in Athens, Greece (Mar 9-13, 2027) at #EACL2027 📅 ARR submission deadline: Aug 6, 2026. Open to all areas of CL/NLP + related fields. Stay tuned for the detailed CfP soon!

How close are we to automating fact-checking with #AI? 🎤🖥️ On August 27, I’ll give an online talk at the AI Academy of @hessianai.bsky.social, covering modern forms of fake news, state-of-the-art systems, and the biggest open challenges. www.linkedin.com/posts/hessia...

Very happy to announce that DEFAME, our state-of-the-art fact-checking system, got accepted to #ICML! 🎉😀 We published the revision on Arxiv, featuring a *new multimodal fact-checking benchmark*, an extensive table showing previous methods, polished figures, and much more: arxiv.org/abs/2412.10510

Über 100 "News"-Seiten verbreiten gezielt Falschinformationen, um die #Bundestagswahl zu manipulieren. Dahinter steckt "Storm-1516", nach #Doppelgänger die nächste #Desinformationskampagne der 🇷🇺 Russen.

🚗 270+ Autos mit Bauschaum im Auspuff - im kompletten Bundesgebiet! Häufig ein Aufkleber der Grünen daneben. Vieles deutet auf eine neue 🇷🇺 russische #Sabotageaktion hin. www.swr.de/swraktuell/b...

Community Notes (die heiße Alternative zu Fact-Checkern) seien "demokratischer" 🤨 www.welt.de/kultur/medie... Willkommen in einer Realität, in der Fakten nicht von Tatsachen, sondern Mehrheiten bestimmt werden.

Eine aktuelle Umfrage des Instituts Allensbach in unserem Auftrag zeigt: #Desinformation und Verschwörungsmythen sind in der deutschen Bevölkerung weit verbreitet – insbesondere bei jungen Menschen und Nutzerinnen und Nutzern von TikTok. Was denkt ihr? www.freiheit.org/de/deutschla...

Oh je, #Desinformation ist unter jungen Menschen deutlich manifestierter als ich bisher angenommen habe 😥 Die erschreckendste neue Zahl: 50% der TikTok-User bezweifeln, dass #Russland gezielt Falschinformationen streue! www.freiheit.org/de/deutschla...

🔎 Prorussische Desinformation nun auch hier auf Bluesky. faktencheck.afp.com/doc.afp.com.... ✉ Noch mehr Falschinfos im Netz gesehen? Hinweise an u.afp.com/whatsapphinweise.

Very unsettling: Over half of young adults view #disinformation as acceptable 🥴 fortune.com/2025/01/19/e... What happens when a world is built on lies? There's no truth, trust, or stability. Then, power simply belongs to the loudest.

You think #deepfakes are the most common type of visual misinformation? Nope, it's using pristine media *out of context*. That is, take a real image/video and caption it with a different story. That kind of #cheapfake makes >40% of visual misinfo, deepfakes only <20%

Doppelgänger has started operating on Bluesky! We're working on mapping out their tactics and techniques to be able to track the operation.

Even almost half a year later, our #fact-checking system "InFact" remains the best text-only fact-checker according to the AVeriTeC leaderboard 💪🏻

Yes, #AI will and already *is* influencing elections. We see #disinformation campaigns (like "Doppelganger") that aim to deform even entire political landscapes 😓 I couldn't agree more with Christian Reuter from @peasec.de. Great article! 👏🏻 www.linkedin.com/posts/activi...

Turns out, #LLMs tend to generate more false or unsupported claims in later sentences of long outputs. So, better double-check when you ask #ChatGPT to write your next essay. 😉 arxiv.org/abs/2411.15993

Darmstadt zählt laut Start-up Monitor 2024 zu den Top 5 Gründungsstandorten in 🇩🇪 und führt Hessen klar an. Mit einer beeindruckenden Dichte von 9,1 Start-ups pro 100.000 Einwohner:innen ist die Stadt ein Zentrum für Innovationen – eingebettet in das dynamische Rhein-Main-Ökosystem.

Universities in 🇩🇪 mass-leaving X. Consequent move, fully supporting the #Xit! 👍🏻 www.zeit.de/digital/2025...

Fact-checkers don’t censor or suppress public debates, they supplement them by weeding out false information to correct misperceptions around false claims. Fact-checking journalism is a crucial safety mechanism, wrote IFCN Director Angie Drobnic Holan in April. www.poynter.org/commentary/2...

A reminder - “ending fact-checking” is an attack on free speech. Fact-checking is publishing information. Abolishing it on your platform is censorship.

#Meta ending professional #fact-checking feels like self-disarming in the fight against growing #disinformation. Both quantity AND quality of #fake-news are escalating. Replacing experts with "Community Notes"? A risky move in an info war! There are better solutions www.wired.com/story/meta-d...

We're happy to announce the achievement of a new milestone in the automation of 🕵🏻‍♂️ claim verification: With our LLM-based fact-checking system, we were able to beat all previous work on 3 benchmarks plus win an international fact-checking challenge! 💪🏻