Advertise on podcast: CERIAS Weekly Security Seminar - Purdue University

Ransomware has evolved from basic digital extortion into a sophisticated, AI-powered threat that's faster,smarter, and more devastating than ever before. In this session, we'll explore how threat actors are weaponizing artificial intelligence to supercharge their operations—from automated reconnaissance and hyper-realistic phishing to malware that adapts in real-time to evade detection. We'll also examine how AI-driven ransomware exploits supply chain vulnerabilities to create cascading disruptions across entire industries.More importantly, we'll discuss practical strategies for fighting back: leveraging AI-powered behavior alanalytics and autonomous response tools, implementing zero-trust architecture,and building true organizational resilience through tested backup and recovery procedures. Whether you're in security operations, incident response, or infrastructure protection, this session will equip you with actionable insights to shift from a prevention-only mindset to one focused on preparedness and rapid recovery in today's evolving threat landscape. About the speaker: Gary Hayslip is an experienced Global Security Executive with a proven track record of delivering innovative security programs that protect billion-dollar enterprises at every touchpoint. He is intensely focused on driving continuous improvement to maximize the efficiency of security programs while minimizing costs. As an insightful thought leader, he possesses strong business acumen and a commitment to organizational mission, values, and goals. He has demonstrated the ability to collaborate with all levels of an organization to champion new ideas, gain buy-in, and build consensus. Hayslip brings extensive experience in information technology, security leadership, physical security, and risk management to his role as the Senior Security Advisor | CISO in Residence for Halcyon.ai. His previous executive positions include multiple roles as Chief Information Security Officer, Chief Information Officer, Deputy Director of IT, and Chief Privacy Officer for the U.S. Navy (Active Duty), the U.S. Navy (Federal Government employee), the City of San Diego, California, Webroot Software, and SoftBank Investments (Vision Fund & Vision Fund II).Hayslip is a proven cybersecurity expert with excellent communication and public speaking skills. He is skilled at explaining complex security and risk concepts to audiences with different levels of knowledge. Hayslip has earned a reputation as a highly effective communicator, author, and keynote speaker. He co-authored the "CISO Desk Reference Guide: A Practical Guide for CISOs – Volumes 1 & 2," "The Executive Primer: An Executive's Guide to Security Programs," "Developing Your Cybersecurity Career Path," and the "The Essential Guide to Cybersecurity for SMBs." He recently coauthored andpublished "Mastering Third Party Risk," a guide aimed specifically for security practitioners to help them manage the risk exposure to organizations from vendors and supply chains. These books are among the top resources for helping CISOs improve their leadership and business skills. Hayslip currently serves as an independent director on several boards and advises various other security and technology firms. He is an active member of the cybersecurity community and belongs to professional organizations such asISC2, NACD, ISACA, and Infragard. Hayslip holds several professional certifications, including CISSP, CISA, and CRISC, and has earned a BS in Information Systems Management from the University of Maryland,University College, and an MBA from San Diego State University.

Annual Security Symposium. Visit: https://ceri.as/2026 Artificial intelligence is rapidly transforming both the opportunities and risks within cybersecurity, creating a new landscape that today's students and researchers will soon inherit and shape. This keynote explores how AI is evolving from a supporting tool to a decision-making system, fundamentally changing how cyber threats are created, detected, and managed. It will examine emerging risks such as deepfakes, model manipulation, and systemic dependencies on shared technologies, while also addressing the growing role of regulation and the challenges of governing systems that are powerful yet often opaque. Most importantly, the session will highlight where the greatest opportunities lie—at the intersection of AI, cybersecurity, and policy—and how the next generation of professionals can play a defining role in building secure, resilient, and trustworthy systems for the future.  About the speaker: Brian J. Peretti is a career member of the Senior Executive Service at the United States Department of the Treasury. In his final position, he served as Treasury's Chief Technology Officer and Deputy Chief Artificial Intelligence (AI) Officer in the Office of Chief Information Officer.As Treasury's Chief Technology Officer, Mr. Peretti establishes, leads, and manages a comprehensive, multi-year strategic and long-range planning process that promotes the vision for IT and ensures consistent progress toward accomplishing the CIO's vision, while identifying and leveraging common technology solutions to support business processes and work methods and/or to improve effectiveness of current technologies while also developing appropriate policy for emerging technology such as Artificial Intelligence, Machine Learning, Biometrics and Quantum Computing. As Treasury's Deputy Chief AI Officer, Mr. Peretti supported Treasury's Chief AI Officer in advancing the Department's deployment of this emerging technology. In this capacity, he oversaw the publication of Treasury's report, Managing Artificial Intelligence-Specific Cybersecurity Risks in the Financial Services Sector, and directed the subsequent lines of effort. Additionally, serving in this position has seen him designated as the Executive Officer for the Department's AI Governance Board as well as the Department's representative to the Office of the Director of National Intelligence's CAIO Council. In addition, Mr. Peretti leads the development of domestic and international operational resilience policy, including cyber, as part of Treasury's Sector Risk Management Agency responsibility for the financial services sector. In this role, he spearheads Treasury's efforts to increase multi-directional sharing of cyber threat and vulnerability information. He also serves as the United States's designated subject matter expert at the Group of 7 Cyber Expert Group (G-7 CEG). Mr. Peretti has served at the Treasury for over 22 years with increasing levels of responsibility, including being named the Senior Career Official Executing the Duties of the Assistant Secretary for Financial Institutions during the transition from the Obama to the Trump Administration. Based on his expertise in critical infrastructure protection and operational resilience, he was detailed to the Department of Homeland Security, Cybersecurity and Infrastructure Security Agency's National Risk Management Center during the intial response to the COVID-19 pandemic and served as the first Senior Advisor for Security and the Economy. He also speadheaded DHS response to the SolarWinds cyber incident. A sought-after speaker and presenter, Mr. Peretti has been the recipient of numerous awards and honors throughout his career. Most recently, he received the 12th Annual Billington CyberSecurity Leadership Award at the 2023 Annual Billington CyberSecurity Summit. Prior to joining the Treasury, Mr. Peretti was an associate in Shook, Hardy & Bacon's Corporate Banking and Finance Section in Washington, D.C., and was the General Counsel for the Wright Patman Congressional Federal Credit Union. He has authored numerous publications related to financial sector operations, including payment systems. Mr. Peretti received his bachelor's degree from Rider University (cum laude) in 1989, and his law degree from American University's Washington College of Law (cum laude) in 1992.

The rapid adoption of mobile applications for managing consumer whole-house battery and energy systems has introduced new questions about software supply chain security. While these applications are not currently integrated with critical infrastructure, their growing role in connected energy environments highlights the importance of understanding the dependencies,permissions, and external services that support their operation. Many of these applications rely on shared third-party libraries, analytics frameworks, and messaging services, creating overlapping software ecosystems across vendors.In this talk, I will present an analysis of several battery-management mobile applications using static and dynamic analysis techniques. The study examines third-party dependencies, Android permission usage, and outbound network activity to identify common software components and shared external infrastructure. The results reveal significant overlap in libraries and permissions across applications, suggesting that vulnerabilities in widely used components could introduce shared risk pathways across multiple vendors. This work highlights the need for stronger dependency governance,permission minimization, and ongoing monitoring as mobile energy applications continue to evolve. About the speaker: Jen Sims is a cybersecurity technical professional in the Cyber Resilience and Intelligence Division at Oak Ridge National Laboratory (ORNL). Her research focuses on resilient cyber-physical systems and vulnerability assessment of technologies used within the electric grid, with particular emphasis on supply chain risk. She also conducts research in cybersecurity for manufacturing and is actively involved in cyber education outreach, engaging students from grade school through graduate programs.Jen earned a Master of Software Engineering and a Bachelor of Computer Science with a concentration in Secure Cyber Systems from the University of Texas at El Paso (UTEP). During her time at UTEP, she founded the Women in Cybersecurity (WiCyS) student chapter and helped launch the university's summer cybersecurity camps.Outside of her research, Jen is passionate about workforce development and cybersecurity education, volunteering with Oak Ridge Computer Science Girls (ORCsGirls) and creating hands-on cybersecurity activities to inspire the next generation of students.

As foundation models, including large language models and multimodal models, are increasingly deployed in complex and high-stakes settings, ensuring their safety has become more important than ever. In this talk, I present a probabilistic perspective on AI safety: safety risks are treated as structured distributions to be discovered and controlled, rather than isolated failures to be patched. I first introduce probabilistic red-teaming methods that characterize distributions of failures, revealing systematic safety risks that standard evaluation often misses. I then describe probabilistic defense methods that control model behavior during deployment by adaptively steering generation toward constraint-aligned distributions. By unifying failure discovery and behavior control under a probabilistic perspective, this talk highlights a distributional approach for understanding and managing safety risks in foundation models. About the speaker: Ruqi Zhang is an Assistant Professor in the Department of Computer Science at Purdue University. Her research focuses on probabilistic machine learning, generative modeling, and trustworthy AI. Prior to joining Purdue, she was a postdoctoral researcher at the Institute for Foundations of Machine Learning (IFML) at the University of Texas at Austin. She received her Ph.D. from Cornell University. Dr. Zhang has been a key organizer of the Symposium on Probabilistic Machine Learning. She has served as an Area Chair and Editor for ML conferences and journals, including ICML, NeurIPS, ICLR, AISTATS, UAI, and TMLR. Her contributions have been recognized with several honors, including AAAI New Faculty Highlights, Amazon Research Award, Spotlight Rising Star in Data Science, Seed for Success Acorn Award, and Ross-Lynn Research Scholar.

Many organizations adopt security frameworks but struggle to turn them into day-to-day operations that reduce risk without slowing delivery. This talk presents a practical operating model that pairs ISO/IEC 27001 (as the certifiable management system that runs governance, risk management, internal audit, and continual improvement) with NIST Cybersecurity Framework 2.0 (as the outcome-focused "dashboard" for aligning security priorities to business objectives and communicating posture to leaders). Attendees will see how to translate business goals into CSF 2.0 current and target profiles, convert those profiles into ISO 27001 objectives and control ownership, and design "evidence by default" workflows that reduce audit fire drills. The session will include real-world design patterns (paved roads, tiered decision rights, exception handling with expiry, and control health metrics) and highlight where assurance programs often drift into "control theater." The goal is a repeatable approach that both practitioners and researchers can critique, improve, and apply. About the speaker: Danny Vukobratovich is a Sr. IT Security Analyst at Purdue University, where he manages Purdue IT's ISO program spanning ISO/IEC 27001 (information security), ISO 9001 (quality management), and ISO/IEC 20000-1 (IT service management). He also oversees Purdue IT's business continuity and disaster recovery planning, with an emphasis on building resilient, auditable operating models that support research and administrative missions. Danny's professional focus is translating risk and governance into practical mechanisms, including clear decision rights, "evidence by design," and metrics that measure control health rather than control presence. His background includes security risk assessments, incident response, monitoring and logging, identity and access management, and standards-based audits across diverse environments. Danny holds the CISSP, ISO/IEC 27001:2022 Lead Implementer, and ITIL 4 Strategic Leader certifications, and an M.S. in Cybersecurity Management.

Deepfake voice technology is rapidly advancing, but how well do current detection systems handle differences in language and writing style? Most existing work focuses on robustness to acoustic variations such as background noise or compression, while largely overlooking how linguistic variation shapes both deepfake generation and detection. Yet language matters: psycholinguistic features such as sentence structure, complexity, and word choice influence how models synthesize speech, which in turn affects how detectors score and flag audio. In this talk,  we will ask questions such as: "If we change the way a person writes, while keeping their voice the same, will a deepfake detector still reach the same decision?" and "Are some text-to-speech and voice cloning models more vulnerable to shifts in writing style than others?" We will then discuss implications for designing robust deepfake voice detectors and for advancing more trustworthy speech AI in an era of increasingly synthetic media. About the speaker:  Thai Le is an Assistant Professor of Computer Science at the Indiana University's Luddy School of Informatics, Computing, and Engineering. He obtained his doctoral degree from the college of Information Science and Technology at the Pennsylvania State University with an Excellent Research Award and a DAAD Fellowship. His research focuses on the trustworthiness of AI/ML models, with a mission to enhance the robustness, safety, and transparency of AI technology in various sociotechnical contexts. Le has published nearly 50 peer-reviewed research works with two best paper presentation awards. He is a pioneer in collecting and investigating so-called text perturbations in the wild, which has been utilized by users and researchers worldwide to study and understand effects of humans' adversarial behaviors on their daily usage with AI/ML models. His works have also been featured in ScienceDaily, DefenseOne, and Engineering and Technology Magazine.

The rise of Industry 4.0 has transformed manufacturing through the integration of cyber-physical systems, connectivity, and real-time data exchange into increasingly automated and intelligent platforms. While these advances improve productivity and efficiency, they also introduce vulnerabilities to cyber-physical attacks that can degrade product quality, damage equipment, and pose safety risks. Effective detection depends on understanding which data sources and levels of granularity provide sufficient visibility for accurate anomaly detection and attack identification. Replicated environments, such as digital twins (DTs), help address the challenges of collecting high-fidelity data and executing complex attack scenarios in live production systems.This talk presents an AI-assisted framework for detecting cyber-physical attacks in smart manufacturing using real machine experimentation complemented by DT–based replication. The framework evaluates multiple data sources, ranging from high-level operational data to low-level control and side-channel signals, to understand how data fidelity and context influence detection performance. A hardware-in-the-loop (HIL) DT is used to replicate machine behavior, safely execute attacks, and enable controlled experimentation that would be impractical in live production environments.Through experiments on a real CNC machining system and its corresponding HIL-based DT, multiple cyber-physical attack scenarios are evaluated using statistical, machine learning, and deep learning-based detection methods. Results demonstrate that detection effectiveness is highly dependent on attack type and data granularity, highlighting the need for domain-aware, multi-source monitoring strategies. The framework is further extended to additive manufacturing, illustrating how insights derived from CNC systems can guide attack detection in related manufacturing domains.Overall, this work demonstrates how combining AI-based detection with real-world experimentation and DT technologies enables more robust and practical security analysis for cyber-physical manufacturing systems. About the speaker: Dr. Bethanie Williams is an R&D, S&E Cybersecurity Engineer at Sandia National Laboratories, where she specializes in applying artificial intelligence (AI) to enhance the security and resilience of cyber-physical systems in critical infrastructure, including power grid systems, healthcare facilities, and advanced manufacturing. She is also actively involved in the Cybersecurity Manufacturing Innovation Institute (CyManII) through her work at Sandia. Bethanie earned her Bachelor of Arts degree as a triple major in Mathematics, Spanish, and Computer Science from Berea College in 2020. During her time at Berea, she was a Bonner Scholar and a member of the women's basketball team, earning All-American honors for her athletic achievements. She completed her Master of Science in Computer Science with a concentration in Cybersecurity at Tennessee Technological University in 2022, under the supervision of Dr. Ambareen Siraj, and earned her Ph.D. in Engineering with a major in Computer Science in 2025 under the guidance of Dr. Muhammad Ismail. Her dissertation, titled "Multi-Source Data Analysis and an Effective AI-Assisted Detection Framework for Cyber-Physical Attacks in Smart Manufacturing," focused on leveraging AI-driven approaches and analyzing various data sources to detect and mitigate cyber-physical attacks in manufacturing systems. Throughout her graduate studies, Bethanie received the College of Engineering Distinguished Fellowship and the National Science Foundation (NSF) Scholarship for Service (SFS). She was a year-round intern at Sandia National Laboratories as part of the Center for Cyber Defenders (CCD) program, where she contributed to national research initiatives under CyManII. Bethanie held several executive leadership roles at Tennessee Tech, including Vice President of Cyber Eagles and Graduate Student Club. She also served as a Ph.D. advisor for Women in Cybersecurity (WiCyS). Through these roles, she actively mentored students, organized outreach events, and fostered a supportive community for women in cybersecurity. Bethanie's current research interests include cyber-physical security, modeling and simulation of industrial control systems, and leveraging AI for advanced manufacturing. As an Early Career R&D, S&E Cybersecurity Engineer at Sandia, she is committed to bridging academic innovation and national security applications to protect critical infrastructure and ensure its resilience.

Parental sharing, sometimes termed "sharenting," refers to ways that parents share information about their children online and is a common mechanism through which young children are exposed to social media. Parental sharing is controversial due to its significant benefits and risks, with researchers highlighting broader concerns regarding long-term implications for children's developing privacy standards. Yet, many parents report a high degree of acceptance for parental sharing, and parents exposing their young children to social media the most are often modeling risky online behaviors. This presentation examines parental sharing in association with privacy and security concepts, research, and interventions toward supporting safe and responsible parental sharing. About the speaker: Mary Jean Amon is a quantitative psychologist focused on human-computer interaction and an Assistant Professor in Indiana University Bloomington's Department of Informatics. Her interdisciplinary research program leverages sensing technologies and advanced analytics to understand and improve dynamic decision-making and performance in the context of complex sociotechnological systems. This includes identifying near-real-time team coordinative patterns that enhance teaming performance, as well as human factors in privacy and security. Amon's quality of work is recognized through publications in top venues, best paper awards, diverse research funding sources, and general dissemination through media outlets like Forbes, New York Times, and Washington Post.

Hardware security is not a new problem, but it is rapidly expanding in both consumer and medical domains due to hyperconnectivity. Medical devices and counterfeit medicines represent a fundamental security challenge. In particular, although counterfeit medicines are not a new issue,the problem continues to worsen as counterfeiting practices become increasingly sophisticated. The counterfeiting of biomedical products poses a serious threat to patient safety, public health, and economic stability in both developed and developing countries, and many current countermeasures remain vulnerable because they provide limited security. In this talk, we will share our work on biomedical hardware security with a focus on pharmaceutical products. We present cyber-physical biomedical security technologies that encode dosage information and authentication into edible biomaterials, enabling serialization, track-and-trace, and authentication at the dosage level. This approach empowers patients to play an active role in combating counterfeit medicines. About the speaker: Young Kim is a professor in the Weldon School of Biomedical Engineering and holds the titles of University Faculty Scholar and Showalter Faculty Scholar at Purdue University. His research centers on co-creating hardware(devices) and software (models) for large-scale societal and healthcare applications. His lab develops hybrid machine learning by combining data analytics with models grounded in optical spectroscopy and light-matter interactions to move beyond big-data, compute-intensive AI and leverage engineers' domain expertise. His work spans optical imaging and spectroscopy, mesoscopic physics, meta materials, cancer research, hardware security, and global health,unified by machine learning and data analytics. His research has been funded by a diverse range of agencies, including NIH, CDC, VA, AFOSR, USAID and Gates Foundation. His primary applications are in global health and rural community health, which address large-scale societal and healthcare challenges in mutually reinforcing ways.

AI-driven machine vision systems are becoming essential in mechanical engineering applications such as fastener classification, yet their increasing connectivity exposes them to adversarial cyberattacks. Model evasion attacks like FGSM can subtly alter input images and cause misclassification, raising concerns about reliability in automated manufacturing.This talk focuses on the role of Explainable AI and human-in-the-loop strategies in detecting and mitigating such attacks. In the presented case study, an EfficientNet-B0 fastener classification model is examined using Grad-CAM visualizations to determine whether shifts inactivation patterns can reveal adversarial manipulation. The study evaluates how FGSM-generated images affect model accuracy and confidence while assessing the XAI system's ability to highlight abnormal regions of attention and the potential for human-in-the-loop approaches to be utilized with XAI techniques as a practical path to strengthening the resilience of AI-based machine vision systems in manufacturing. About the speaker: Dr. Vijayanth Tummala is a Researcher in Cybersecurity and Human-AI Interaction. His research spans artificial intelligence and cybersecurity across interdisciplinary areas, including AI and Cybersecurity leadership, AI literacy, and computer vision applications. He was one of only seven recipients to receive the Best Paper Award in the AI track at ASME's IMECE conference held in November 2024, which features over 2,400 submissions annually. Previously, he held key leadership roles, including leading the NSA CAE-CD designation, launching graduate programs as part of a $1.5 million EDA grant received by his previous employer, and partnering with the Allen County High-Tech Crimes Unit.

Collaborative robots and machine-learning-based virtual agents are increasingly entering the human workspace with the aim of increasing productivity, enhancing safety, and improving the quality of our lives. These agents will dynamically interact with a wide variety of people in dynamic and novel contexts, increasing the prevalence of human-machine teams in applications spanning from healthcare and manufacturing to household assistance. My research aims to create transparent embodied systems that can support users and interact with humans, pushing the frontier of real-world robotics systems towards those that understand human behavior, maintain interpretability, and coordinate with high performance.  In this talk, I will cover a set of works that enable robots to 1) understand and learn from diverse human users, 2)  learn interpretable, human-readable tree-based control policies directly via reinforcement learning, and 3) provide users with information online to improve situational awareness and facilitate effective human-robot collaboration. About the speaker: Dr. Rohan Paleja is an Assistant Professor in the Department of Computer Science at Purdue University. He directs the Strategies for Collaboration, Autonomy, Learning, and Exploration in Robotics Lab. The SCALE Robotics Lab focuses on advancing machine learning and artificial intelligence to improve robot learning, human-robot interaction, and multi-agent collaboration. Their goal is to equip autonomous agents with the ability to operate in the diverse, unstructured, and human-rich environments these agents will encounter in the real world.Dr. Paleja's research interests cover a broad range of topics, namely Explainable AI (xAI), Interactive Robot Learning, and Multi-Agent Collaboration. Prior to Purdue, Dr. Paleja was a Technical Staff Researcher in the Artificial Intelligence Technology group at MIT Lincoln Laboratory, where he collaborated with the Air Force Experimental Operations Unit and the Army Research Lab. Prior to that, he earned his Ph.D. in Robotics at the Georgia Institute of Technology in 2023.His work has received multiple awards, including a Best Paper Finalist Award at the Conference of Robot Learning (CoRL) and a Best Workshop Paper Award at the International Conference of Computer Vision (ICCV) Multi-Agent Relational Reasoning Workshop.

Over the past several years, CL0P has executed multiple mass exploitation campaigns using zero-day vulnerabilities in popular software products that resulted in mass data exfiltration. In this talk we'll take a look at the vulnerabilities that enabled their access, discuss ways defenders could have detected the exploits, and explore hardening recommendations to make public facing applications harder to compromise. About the speaker: Peter Ukhanov is a Principal Consultant with the Google Public Sector (Mandiant) IR team. Prior to joining Mandiant, Peter worked at Dragos focusing on OT/ICS environments. He started his career in incident response and digital forensics in 2014 at the Defense Information Systems Agency, spending almost 7 years supporting various Department of Defense entities.

In this talk, I will discuss recent research projects at the intersection of software security and automated reasoning. Specifically, I will present our work on assessing the exploitability of the Android kernel and developing complex exploits for it, as well as our efforts to uncover bugs in Rust's unsafe code through fuzzing.Throughout the talk, I will highlight how Large Language Models (LLMs) can support both attackers and defenders in analyzing complex software systems, and I will present key lessons on using LLMs effectively along with the practical challenges that arise when integrating them into software security workflows. About the speaker: Dr. Antonio Bianchi's research interest lies in the area of Computer Security. His primary focus is in the field of security of mobile devices. Most recently, he started exploring the security issues posed by IoT devices and their interaction with mobile applications. As a core member of the Shellphish and OOO teams, he played and organized many security competitions (CTFs), and won the third place at the DARPA Cyber Grand Challenge.

Online video games have evolved into vast financial ecosystems where real and virtual value mix at scale. This presentation shows how these spaces serve as efficient laundering channels, converting illicit funds from organized crime, sanctions evasion, terrorist financing, and digital fraud into assets that appear legitimate. Illicit value typically enters via card not present transactions, stolen digital wallets, and scam revenues before it is routed into platform marketplaces. From there, funds convert into tradeable virtual assets such as cosmetics, currencies, loot boxes, and content bundles, which can be divided into thousands of rapid microtransactions. Widely cited estimates place illicit financial flows at 2 to 5 percent of global GDP (roughly $800 billion to $2 trillion a year), while in game spending will reach $74.4 billion in 2025, providing liquidity, speed, and plausible deniability. About the speaker: Stephen Flowerday is a Professor in the School of Computer and Cyber Sciences at Augusta University. His research focuses on cybersecurity management, cybercrime, behavioral information security, and human-centric cybersecurity at the intersection of technology, processes, and people. His work has been supported by IBM, THRIP, the NRF, SASUF, Erasmus, and GMRDC. He serves as an associate editor and frequent reviewer for leading journals and conferences, and has reviewed grants for the Israeli NSF, the South African NRF, the U.S. NSF, and Bahrain's DHE.

Reasoning—the process of drawing conclusions from prior knowledge—is a hallmark of intelligence. Large language models, and more recently, large reasoning models have demonstrated impressive results on many reasoning-intensive benchmarks. Careful studies over the past few years have revealed that LLMs may exhibit some reasoning behavior, and larger models tend to do better on reasoning tasks. However, even the largest current models still struggle on various kinds of reasoning problems. In this talk, we will try to address the question: Are the observed reasoning limitations of LLMs fundamental in nature? Or will they be resolved by further increasing the size and data of these models, or by better techniques for training them? I will describe recent work to tackle this question from several different angles. The answer to this question will help us to better understand the risks posed by future LLMs as vast resources continue to be invested in their development. About the speaker: Abulhair Saparov is an Assistant Professor of Computer Science at Purdue University. His research focuses on applications of statistical machine learning to natural language processing, natural language understanding, and reasoning. His recent work closely examines the reasoning capacity of large language models, identifying fundamental limitations, and developing new methods and tools to address or workaround those limitations. He has also explored the use of symbolic and neurosymbolic methods to both understand and improve the reasoning capabilities of AI models. He is also broadly interested in other applications of statistical machine learning, such as to the natural sciences.