Exposing influence campaigns in the age of LLMs: a behavioral-based AI approach to detecting state-sponsored trolls

The detection of state-sponsored trolls operating in influence campaigns on social media is a critical and unsolved challenge for the research community, which has significant implications beyond the online realm. To address this challenge, we propose a new AI-based solution that identifies troll accounts solely through behavioral cues associated with their sequences of sharing activity, encompassing both their actions and the feedback they receive from others. Our approach does not incorporate any textual content shared and consists of two steps: First, we leverage an LSTM-based classifier to determine whether account sequences belong to a state-sponsored troll or an organic, legitimate user. Second, we employ the classified sequences to calculate a metric named the "Troll Score", quantifying the degree to which an account exhibits troll-like behavior. To assess the effectiveness of our method, we examine its performance in the context of the 2016 Russian interference campaign during the U.S. Presidential election. Our experiments yield compelling results, demonstrating that our approach can identify account sequences with an AUC close to 99% and accurately differentiate between Russian trolls and organic users with an AUC of 91%. Notably, our behavioral-based approach holds a significant advantage in the ever-evolving landscape, where textual and linguistic properties can be easily mimicked by Large Language Models (LLMs): In contrast to existing language-based techniques, it relies on more challenging-to-replicate behavioral cues, ensuring greater resilience in identifying influence campaigns, especially given the potential increase in the usage of LLMs for generating inauthentic content. Finally, we assessed the generalizability of our solution to various entities driving different information operations and found promising results that will guide future research.

[Computerized tomography and magnetic resonance in the diagnosis of recurrent rectal neoplasms: comparison of reliability and errors of both methods]

Our work was aimed at assessing the accuracy of CT and MRI in the early identification of postoperative recurrences of rectosigmoid cancer, quantifying false positive and false negative ratios and questionable findings. A homogeneous series of 50 patients submitted to surgery for primary rectosigmoid carcinoma was studied with both CT and MRI and followed-up for up to 2 years: local recurrences were observed in 15 patients (30%) which occurred within the first year of surgery in 67% of cases and were in extraluminal site in 86% of cases. CT appeared to be a reliable and highly sensitive screening method, with 82% sensitivity: only 2 false negatives were observed, which were nevertheless followed-up because CT had yielded questionable, and not negative, findings. In contrast, CT had only 78% specificity, being unable to differentiate fibrosis, displaced normal structures and recurrences in 7 cases; its positive predictive value (PPV) was 66%, with 89% negative predictive value (NPV) and 80% accuracy. MRI had a complementary role to CT, because of its capabilities in discriminating all the questionable CT cases and in identifying all CT false positives, thanks to its higher specificity (100%). MRI had 74% sensitivity, which was lower than that of CT; MRI yielded 2 false negatives which, however, had been previously diagnosed with CT. MRI had 100% PPV, 89% NPV and 92% accuracy. Therefore, for the early detection of rectal cancer recurrences, the following diagnostic protocol is suggested: CT should be performed first, as a screening method, within 2-4 months of surgery, and repeated every 6-8 months during the first 2 years-together with CEA values monitoring. MRI should be reserved to the patients in whom CT findings were positive, questionable, or in disagreement with clinical symptoms and/or with increasing CEA values. If MRI fails to solve the diagnostic doubt, a CT-guided biopsy of the mass should be performed.