DR. MATTHEW CANHAM
AI Manipulation of Humans & Cognitive Defense | 3x Black Hat USA Speaker
Expert Witness | Former FBI | PhD | Executive Director, CSI
RESEARCH PAPERS
2026
Bryce-Allen Bagley, Nathaniel Rose, Quintus Kilbourn, Matthew Canham
Brain-computer interfaces (BCIs) are a class of diverse hardware modalities, associated software, and connected devices which are widely used in a variety of fields, including neurosurgery, biomedical data analysis, and neuroimaging. Recent years have seen rapid advancements in BCI technology, and neurotechnology more broadly, with the first devices now passing clinical trials, early examples of consumer hardware entering the market, and many variants of consumer and medical hardware with increasingly extensive capabilities being developed rapidly. However, research and development in security for BCIs--known as neurosecurity--lags significantly behind the capabilities of BCIs themselves. In an effort to address as many vulnerabilities as feasible immediately, in this paper we review the current state of the art in neurosecurity, thoroughly survey the breadth and complexity of both firmly established and highly probable security threats to BCI systems, and provide recommendations of existing methods from cybersecurity, hardware security, and machine learning which can immediately be applied to address some of these gaps in neurosecurity.
Anuradha Rangarajan Julie Haney, Matthew Canham, Mike Elkins, Lisa Flynn, Matthew Gordin, Victoria Granova, Wenjing Huang, Jody Jacobs, Greg Moody, Michael Ross, Robert Thomson, Joe Uchill
We present the key themes from the workshop “ConnectCon: Minding the Gaps in Human-Centered Cybersecurity,” which can inform efforts of researchers and practitioners toward a collective goal of improving cybersecurity outcomes for all.
2025
Julie Haney, Matthew Canham, Mike Elkins, Lisa Flynn, Matthew Gordin, Victoria Granova, Wenjing Huang, Jody Jacobs, Greg Moody, Ann Rangarajan
We present the key themes from the workshop “ConnectCon: Minding the Gaps in Human-Centered Cybersecurity,” which can inform efforts of researchers and practitioners toward a collective goal of improving cybersecurity outcomes for all.
Chad C Tossell, Cara A Spencer, Mica R Endsley, Matthew Canham, Laura Steckman, Allison Hayman, Leanne Hirshfield
For the second year in a row, the World Economic Forum’s Global Risks Report identifies propaganda, deception, and contested information as top threats to global stability. These information threats are no longer isolated events; they are persistent, evolving, and amplified by emerging technologies such as AI. While technical defenses remain critical, they are not sufficient. The human element, including how individuals perceive, interpret, and respond to information, has become central to understanding and addressing these challenges. This panel brings together leading experts from defense, academia, and industry to examine cognitive security as a growing frontier for the human factors community. Panelists will explore the psychological and contextual mechanisms that make people and systems susceptible, share innovative interventions grounded in HF principles, and highlight ways HF professionals can contribute to building cognitive resilience.
Yvette Butler, Chibundo Egwuatu, Matthew Canham, Ben D Sawyer
Sex worker artificial intelligence (AI) replicas are boutique Human Digital Twins (HDTs) whose unique selling point is their ability to simulate conversation with a real person whose personality is replicated through AI. Notably, they present a model for sex worker–AI collaboration that protects the intellectual property (IP) rights of the original worker, potentially increasing their earnings rather than diminishing them. This article challenges assumptions that AI will work as a pristine weapon to be wielded against sex workers and their livelihoods. Instead, we explore AI’s potential as an asset to protect sex workers’ status as agents and argue for the substantial inclusion of sex workers as thought partners in the future of AI replica technology and legal responses to IP protections threatened by the rise of AI in the USA. This article considers how the sex workers’ response to the rising challenge of AI can influence future IP, AI, and sex tech thought.
Aileen Fezzie, Rosanna E Guadagno, Matthew Canham
Online manipulation and cognitive attacks such as misinformation and foreign influence exploit language to shape perception and persuasion. The theory of linguistic relativity suggests that language structures cognition, influencing concepts like color and time. Research indicates that linguistic differences impact trust and decision-making, with speakers of weak future-time-reference (FTR) languages (e.g., Mandarin) displaying different investment behaviors than strong-FTR speakers (e.g., English). Additionally, cultural orientations, such as individualism and collectivism, influence susceptibility to manipulation tactics like commitment, consistency, and social proof. The language of a message may activate cultural schemas, altering its effectiveness, particularly for bilingual individuals. This has critical implications for countering cognitive attacks, suggesting that tailoring messaging to align with linguistic and cultural frameworks can enhance persuasive impact. More research is needed to understand how linguistic delivery moderates influence, informing strategies to resist persuasion in multilingual environments.
Torvald Ask, Stefan Sütterlin, Lea Müller, Ricardo Lugo, Dominic Saari, Hilkka Grahn, Matthew Canham, Daniel Hermansen, Benjamin Knox
Advances in science and technologyhave expanded how cognition can be targeted, manipulated, and weaponized. These developments pose several challenges to international security and have heightened the need for a coherent framework of Cognitive Security. We propose a unifying definition of Cognitive Security as “the state of having trusted boundaries protecting cognitive assets against all forms of unauthorized access and interference”.Cognitive assets are defined broadly, encompassing individual neurocognitive processes, organizationaland internationaldecision-making structures, and artificial intelligence systems. We identify four interdependent cognitive resilience factors essential to maintaining Cognitive Security: (1) cognitive agility, the ability to adaptively manage information processing and decision processes; (2) machine psychology, safeguarding cognition-like functions in machine learningsystems; (3) neurosecurity, protecting neural systems from biological or technological intrusion; and (4) systems engineering, the design oforganizational and societal structures that foster cognitive resilience. We argue that effective Cognitive Security requires a holistic approach that integrates these factors. By consolidating fragmented perspectives into a holisticframework, this paper advances Cognitive Security as a distinct discipline and provides a foundation forinternationalresearch, policy, and practice aimed at ensuringcognitive,and ultimately,international security.
Robert H Thomson, Quan Nguyen, Essien Ayanam, Matthew Canham, Thomas C Schmidt, Matthias Wählisch, Eric Osterweil
Humanity is currently facing an existential crisis about the nature of truth and reality driven by the availability of information online which overloads and overwhelms our cognitive capabilities, which we call Cyber-Psychosis. The results of this Cyber-Psychosis include the decline of critical thinking coupled with deceptive influences on the Internet which have become so prolific that they are challenging our ability to form a shared understanding of reality in either the digital or physical world. Fundamental to mending our fractured digital universe is establishing the ability to know where a digital object (i.e. a piece of information like text, audio, or video) came from, whether it was modified, what it is derived from, where it has been circulated, and what (if any) lifetime that information should have. Furthermore, we argue that on-by-default object security for genuine objects will provide the necessary grounding to support critical thinking and rational online behavior, even with the ubiquity of deceptive content. To this end, we propose that the Internet needs an object security service layer. This proposition may not be as distant as it may first seem. Through an examination of several venerable (and new) protocols, we show how pieces of this problem have already been addressed. While interdisciplinary research will be key to properly crafting the architectural changes needed, here we propose an approach for how we can already use fallow protections to begin turning the tide of this emerging Cyber-Psychosis today!
2024
Matthew Canham, Shanée Dawkins, Jody Jacobs
This study investigated factors influencing repeat clicking behavior in phishing attacks, focusing on individual characteristics like personality traits and expertise, social engineering tactics using the NIST Phish Scale, and contextual influences such as world events. Participants were surveyed on their individual differences and exposed to multiple phishing email campaigns to assess their susceptibility.
2023
Matthew Canham
The study identifies two distinct employee groups regarding phishing susceptibility: repeat clickers, who frequently fall for phishing attacks, and protective stewards, who reliably detect and report phishing attempts. Factors like confidence in email judgment and checking habits differ between these groups, suggesting varying levels of vulnerability to cyber threats.
Torvald F Ask, Ricardo G Lugo, Stefan Sütterlin, Matthew Canham, Daniel Hermansen, Benjamin J Knox
Cognitive Warfare takes advantage of novel developments in technology and science to influence how target populations think and act. Establishing adequate defense against Cognitive Warfare requires examination of modus operandi to understand this emerging action space. This includes the goals and methods that can be realized through science and technology. Recent literature suggests that both human and nonhuman cognition should be considered as targets of Cognitive Warfare. There are currently no frameworks allowing for a unified way of conceptualizing short-term and long-term Cognitive Warfare goals and attack methods that are domain- and species-agnostic. There is a need for a framework developed through a bottom-up approach that is informed by neuroscientific principles to capture relevant aspects of cognition. The framework should be at a level of complexity that is actionable to decision-makers in war. In this paper, we attempt to cover the existing gap by proposing the Unplug, Corrupt, disOrganize, Diagnose, Enhance (UnCODE) system for classifying the goals and methods of Cognitive Warfare. The system is neurocentric by conceptualizing Cognitive Warfare goals from the perspective of how adversarial methods relate to neural information processing in an individual or society. The UnCODE system identifies five main classes of goals: 1) Eliminating a target’s ability to produce outputs, 2) degrading a target’s capacity to process inputs and produce outputs, 3) biasing a target’s input-output activity, 4) monitoring and understanding the input-output relationships in targets, and 5) enhancing a target’s capacity and ability to process inputs and produce outputs. Methods can be divided in two categories based on access to the target’s neural system: direct access and indirect access. The UnCODE system is domain- and species-agnostic and allows for interdisciplinary commensurability when communicating attack paths across domains. In sum, the UnCODE system is a unifying framework that captures how multiple methods can be used to reach the same Cognitive Warfare goals.
2022
Matthew Canham, Juliet Tuthill
By nearly every metric, the status quo of information security is not working. The interaction matrix of attacker-defender dynamics strongly favors the attacker who only needs to be lucky once. We argue that employing social engineering active defense (SEAD) will be more effective to countering malicious actors than maintainingthe traditional passive defensive strategy. The Offensive Countermeasures (OCM) approach to defense advocates for three categories of countermeasures: annoyance, attribution, and attack. Annoyance aims to waste the attacker’s time and resources with the objective of not only deterrence but also to increase the probability of detection and attribution. Attribution attempts to identify who is launching the attack. Gathering as much threat intelligence on who the attacker is, provides the best possible defense against future attacks. Finally, attack involves running code on the attacker’s system for the purpose of deterrence and attribution. In this work, we advocate for utilizing similar approaches to deny, degrade, and de-anonymize malicious actors by using social engineering tools, tactics, and procedures against the attackers. Rather than fearing the threats posed by synthetic media, cyber defenders should embrace these capabilities by turning theseagainst criminals. Future research should explore ways to implement synthetic media and automated SEAD methods to degrade the capabilities of online malicious actors.
Matthew Canham, Clay Posey, Michael Constantino
To better understand employees’ reporting behaviors in relation to phishing emails, we gamified the phishing security awareness training process by creating and conducting a month-long “Phish Derby” competition at a large university in the U.S. The university’s Information Security Office challenged employees to prove they could detect phishing emails as part of the simulated phishing program currently in place. Employees volunteered to compete for prizes during this special event and were instructed to report suspicious emails as potential phishing attacks. Prior to the beginning of the competition, we collected demographics and data related to the concepts central to two theoretical foundations: the Big Five personality traits and goal orientation theory. We found several notable relationships between demographic variables and Phish Derby performance, which was operationalized from the number of phishing attacks reported and employee report speed. Several key findings emerged, including past performance on simulated phishing campaigns positively predicted Phish Derby performance; older participants performed better than their younger colleagues, but more educated participants performed poorer; and individuals who used a mix of PCs and Macs at work performed worse than those using a single platform. We also found that two of the Big Five personality dimensions, extraversion and agreeableness, were both associated with poorer performance in phishing detection and reporting. Likewise, individuals who were driven to perform well in the Phish Derby because they desired to learn from the experience (i.e., learning goal orientation) performed at a lower level than those driven by other goals. Interestingly, self-reported levels of computer skill and the perceived ability to detect phishing messages failed to exhibit a significant relationship with Phish Derby performance. We discuss these findings and describe how focusing on motivating the good in employee cyber behaviors is a necessary yet too often overlooked component in organizations whose training cyber cultures are rooted in employee click rates alone.
Matthew Canham, Stefan Sütterlin, Torvald F Ask, Benjamin J Knox, Lauren Glenister, Ricardo G Lugo
Humans quickly and effortlessly impose context onto ambiguous stimuli, as demonstrated through psychological projective testing and ambiguous figures. This feature of human cognition may be weaponized as part of an information operation. Such Ambiguous Self-Induced Disinformation (ASID) attacks would employ the following elements: the introduction of a culturally consistent narrative, the presence of ambiguous stimuli, the motivation for hypervigilance, and a social network. ASID attacks represent a low-risk, low-investment tactic for adversaries with the potential for significant reward, making this an attractive option for information operations within the context of grey-zone conflicts.
Stefan Sütterlin, Benjamin J Knox, Kaie Maennel, Matthew Canham, Ricardo G Lugo
Digitalization in the health sector is, as in all societal domains, motivated by a range of anticipated positive consequences, such as increased effectiveness of prevention, treatment and follow-up, a generally improved resource efficiency and improved health care availability. This chapter will discuss how the ambition of achieving sustainable health goals may be affected by measures of digitalization. This will be done by covering digitalization from a cyber security perspective and how new potential threats to privacy may influence the public’s trust in their health care system, thereby affecting the envisaged goals of sustainable health care performance. It will also discuss further how digitalization in the healthcare sector unleashes an enormous potential in terms of cost-effectiveness, decentralization and the availability of specialist services and expertise, which risks and countermeasures these changes entail, and how they are currently dealt with and the role of cyber resilience in ensuring rapid digitalization does not come at the cost of essential trust mechanisms that are quid pro quo in the health sector.
2021
Matthew Canham, Clay Posey, Delainey Strickland, Michael Constantino
Organizational cybersecurity efforts depend largely on the employees who reside within organizational walls. These individuals are central to the effectiveness of organizational actions to protect sensitive assets, and research has shown that they can be detrimental (e.g., sabotage and computer abuse) as well as beneficial (e.g., protective motivated behaviors) to their organizations. One major context where employees affect their organizations is phishing via email systems, which is a common attack vector used by external actors to penetrate organizational networks, steal employee credentials, and create other forms of harm. In analyzing the behavior of more than 6,000 employees at a large university in the Southeast United States during 20 mock phishing campaigns over a 19-month period, this research effort makes several contributions. First, employees’ negative behaviors like clicking links and then entering data are evaluated alongside the positive behaviors of reporting the suspected phishing attempts to the proper organizational representatives. The analysis displays evidence of both repeat clicker and repeat reporter phenomena and their frequency and Pareto distributions across the study time frame. Second, we find that employees can be categorized according to one of the four unique clusters with respect to their behavioral responses to phishing attacks—“Gaffes,” “Beacons,” “Spectators,” and “Gushers.” While each of the clusters exhibits some level of phishing failures and reports, significant variation exists among the employee classifications. Our findings are helpful in driving a new and more holistic stream of research in the realm of all forms of employee responses to phishing attacks, and we provide avenues for such future research.
Matthew Canham
How do you know that you are actually talking to the person you think you are talking to? Deepfake and related synthetic media technologies may represent the greatest revolution in social engineering capabilities yet developed. In recent years, scammers have used synthetic audio in vishing attacks to impersonate executives to convince employees to wire funds to unauthorized accounts. In March 2021, the FBI warned the security community to expect a significant increase in synthetic media enabled scams over the next 18 months. The security community is at a highly dynamic moment in history in which the world is transitioning away from being able to trust what we experience with our own eyes and ears. This presentation proposes the Synthetic Media Attack Framework to describe these attacks and offer some easy to implement, human-centric countermeasures. This framework utilizes five dimensions: Medium (text, audio, video, or a combination), Interactivity (pre-recorded, high asynchrony, low asynchrony, or real-time), Control (human puppeteer, software, or a hybrid), Familiarity (unfamiliar, familiar, close), and Intended Target (human or automation, an individual target, or a broader audience), to describe synthetic media social engineering attacks. While several technology-based methods to detect synthetic media such as currently exist, this work focuses discussion on human centered countermeasures to synthetic media attacks because most technology-based solutions are not readily available to the average user and are difficult to apply in real-time. Effective security policies can help users spot inconsistencies between the behaviors of a legitimate actor and a syn-puppet. Proof-of-life statements will effectively counter most virtual kidnappings leveraging synthetic media. Significant financial transfers should require either multi-factor authentication (MFA) or multi-person authorization. These ‘old-school’ solutions will find new life in the emerging world of synthetic media attacks.
Ben D Sawyer, Dave B Miller, Matthew Canham, Waldemar Karwowski
In developing the highly useful technologies, knowledge from human factors and ergonomics (HF/E) can be of great use, especially to designers charged with the difficult task of dovetailing humans and machines in complex systems built to navigate sometimes chaotic environments. The role of HF/E in A 3 design remains centered around the goal that A 3 self‐action is ideal to provide maximum benefit to humans while increasing the likelihood of task success. This chapter is written between “artificial intelligence (AI) winters,” times of decreased funding of AI technologies, indeed at a time of great optimism and investments in A 3 technologies. It argues that design is key to all A 3 technologies, and that the advent of autonomy which has no need for humans is not …
Stefan Sütterlin, Torvald F Ask, Sophia Mägerle, Sandra Glöckler, Leandra Wolf, Julian Schray, Alava Chandi, Teodora Bursac, Ali Khodabakhsh, Benjamin J Knox, Matthew Canham, Ricardo G Lugo
AI-generated “deep fakes” is increasingly used by cybercriminals conducting targeted and tailored social engineering attacks, and for influencing public opinion. To raise awareness and efficiently train individuals in recognizing deep fakes, understanding individual differences in the ability to recognize them is central. Previous research suggested a close relationship between political attitudes and top-down perceptual and cognitive processing styles. In this study, we investigate the impact of political attitudes and agreement with the political message content on individual deep fake recognition skills. 163 adults (72 females = 44.2%) judged a series of video clips with politicians’ statements across the political spectrum regarding their authenticity and their agreement with the message content. Half of the presented videos were fabricated via lipsync technology. In addition to agreement with each statement made, global political attitudes towards social and economic topics were assessed via the Social and Economic Conservatism Scale (SECS). There were robust negative associations between participants’ general and social conservatism and their ability to recognize fabricated videos, especially when where there was agreement with the message content. Deep fakes watched on mobile phones and tablets were considerably less likely to be recognized compared to when watched on stationary computers. This is the first study to investigate and establish the association between political attitudes and interindividual differences in deep fake recognition. The study supports recently published research suggesting relationships between conservatism and perceived credibility of conspiracy theories and fake news in general. Implications for further research are discussed.