The NDSS’25 statistics provide a stark, quantitative snapshot of a research community at a critical inflection point, defined by two key trends: explosive submission growth and a seismic shift in research focus.
The most striking figure is the doubling of submissions from 2024 (694 to 1311). This exponential increase, particularly in the Fall cycle (946 submissions), signifies a healthy, expanding field and represents substantial challenges.
The burden on the 167 Program Committee (PC) members was immense, as evidenced by the 3,629 reviews (on average, more than 20 reviews per TPC member) and nearly 9,600 discussion comments. This data indicates “reviewer fatigue,” a concern across top-tier conferences.
We can only speculate about the reasons for the immense increase in paper submissions. Still, we are aware that the NDSS Symposium was upgraded as a top-tier venue in Chinese rankings this year, which seems to have led to a substantial increase in submissions from Chinese authors. Regarding accepted papers, authors from the US and China were at par this year (97 vs. 96 accepted papers).
The second major insight comes from the topic distribution. “ML/AI/LLM Security” has emerged as the most dominant submission topic. This demonstrates the community’s rapid and necessary pivot to address the security and privacy implications of the generative AI (GenAI) revolution. While it is the most popular field, its acceptance rate is not the highest, suggesting a highly competitive and perhaps noisy area with a mix of quality, which may be common for “hot” new topics.
Pressing Issues for the Community
This unprecedented growth, coupled with the new “hot topic,” creates significant challenges for the integrity and sustainability of the academic process.
The Sustainability of Peer Review
The NDSS’25 numbers show that the traditional peer review model is under heavy strain. A PC of a similar size cannot absorb a 100% increase in submissions without a proportional decrease in review quality or an untenable increase in reviewer workload. This growth risks reviewer burnout, rushed evaluations, and a potential inability to separate novel contributions from incremental work. The community must grapple with this scaling challenge through new review models, stricter submission guidelines, or larger, more incentivized review pools — likely all three.
The “AI in the Loop” Dilemma
The dominance of AI as a topic is mirrored by its emergence as a tool—a duality that presents a profound challenge. The use of AI in paper writing and reviewing is a new and pressing issue.
- AI-Assisted Writing: While LLMs can help non-native speakers or streamline drafting, they also risk homogenizing academic writing, obscuring a lack of novel ideas with polished prose, and introducing subtle, unverified “hallucinations” as fact. Reviewers have accused authors of AI involvement in paper writing several times, requiring individual case-by-case assessments by the TPC chairs (or an Ethics or future dedicated AI subcommittee).
- AI-Assisted Reviewing: Just as we say reviewers accuse authors of GenAI usage for paper writing, authors accuse reviewers of GenAI usage for their reviews. This is an even greater ethical minefield. Using AI to summarize or draft reviews breaches confidentiality—a cornerstone of peer review—by exposing unpublished work to a third-party model. It also encourages superficial, “checklist-style” reviews that may miss deep methodological flaws or novel insights.
The TPC chairs of the NDSS Symposium and similar conferences must deal with and solve individual accusations and cases on both sides. The community needs to develop clear guidelines from authors and reviewers on which use of GenAI is acceptable and which is not.
In essence, the NDSS ’25 statistics are a microcosm of the security field: It is more popular and relevant than ever, but it is also struggling to manage its success. The community is tasked with securing the AI models that are simultaneously flooding its submission systems and threatening to upend its foundational peer review processes.
Christina Pöpper is Associate Professor of Computer Science at New York University Abu Dhabi and Program Co-Chair of NDSS Symposium 2025.