E-PhishGEN: Unlocking Novel Research in Phishing Email Detection
Workshop ACM Workshop on Artificial Intelligence Security [BEST PAPER AWARD]
Taipei, Taiwan
Oneliner: Most research in phishing email detection uses outdated datasets, so we try to make things a bit better.
First time in Taiwan, first time in Taipei. In the last 18 months, I have been in Asia 4 times (Singapore, Japan, Vietnam, Taiwan), all for “work-related” reasons. The beauty of this “job”.
CCS’25 was really nice. I liked it more than CCS’24. The venue was much larger and there was more breathing room—whereas the previous year everything felt very cramped. However, the social dinner at CCS’24, as well as the food in general, was much better: I love Asian food, but I didn’t particularly enjoy anything served during the meals at CCS’25.
Nevertheless, I chaired one session at CCS’25, and it was amazing. The talks were great and the papers were very diverse, covering RAG, privacy, poisoning, test-time attacks against “production” ML-based detectors, LLMs, and even certifiable robustness of… _k_NN (because ML is not just about neural networks!). Props to all the speakers!
Finally, as for the reason I attended CCS’25, it was to present a paper I co-authored at AISec’25. The paper had been selected as a spotlight: in my career, I had never seen such positive reviews (2xStrong Accept, 1xAccept, 1xWeak Accept) and we were all surprised (and happy!) of how well the paper was appreciated by the reviewers. We have done all we could to further improve it, and I tried to convey its takeaways during the 5-minute spotlight talk. Yet, after a very insightful poster session, I would have never expected that our paper would win the Best Paper Award! I think the other papers were great too—as were many of those for which I had the opportunity to talk with their co-authors during the poster session.
Still, none of this would have been possible without my great co-authors, especially Stefan, Luca, and Eugenio!