-: AI & Digital Forensics - Review of Baltezarević et. al. (2026)

Overview of the Baltezarević et al (2026) Report

This report presents guidelines for digital forensics examinersʼ use of artificial intelligence (AI), based on the publication: Enhancing Human-Machine Cooperation through the Integration of Empathetic AI into Corporate Cybersecurity and Forensic Operations, by Baltezarević et al. (2026). Their report emphasizes oversight, appropriate analytical assistance, protection of sensitive information, and preservation of examinersʼ decision-making abilities.

Purpose of the Baltezarevic et. al. Research

The authors explain that their article “examines the concept of empathetic artificial intelligence and looks at how it can improve collaboration between humans and this new technology”. Its particular focus is “corporate cybersecurity and forensic investigations, and how empathetic AI can improve their processes” (Baltezarević et al., 2026, p. 374).

The research addresses three questions:

Applications: “how empathic AI systems are being utilized in fields like corporate cybersecurity, forensic operations, and human-machine interactions” (p. 374).
Potential benefits: whether integration could “strengthen trust, improve decision-making, and promote collaboration during challenging investigations and operations” (p. 374).
Implementation challenges: the “key technical, ethical, and operational hurdles” associated with implementation and possible responses “without sacrificing that crucial human oversight” p. 374).

Scope of the Research

The authors describe their work as a review that “brings together insights from psychology, human-computer interaction, artificial intelligence, and cybersecurity” (p. 374). Accordingly, the recommendations below are limited to the articleʼs discussion of AI-assisted forensic analysis, empathetic AI, human–machine collaboration, and related cybersecurity operations.

The researchʼs maturity limits its practical application. The authors acknowledge that “most research on empathic artificial intelligence in cybersecurity is currently theoretical or based only on simulations” (p. 379). The following entries therefore distinguish proposed uses and recommendations from implementation problems identified in the article. All page citations refer to the articleʼs printed journal pagination.

Recommendations and Findings

Use AI to assist with evidence organization, anomaly identification, and threat tracing. The authors identify appropriate assistance as “organizing evidence, identifying anomalies, and tracing the origins of threats,” accomplished by “integrating machine learning with human intuition” (p. 374). These are the articleʼs expressly identified applications for digital forensic analysts.

Preserve the examinerʼs contextual and legal contribution. AI-assisted data processing should remain paired with human interpretation. The authors state that “AI can handle huge amounts of forensic data much more effectively than traditional methods, while human analysts bring in the necessary context and legal insights” (p. 374).

Define when AI may assist and when humans must take control. The article supplies an explicit governance guideline: “Humans and machines must have clear roles, and governance frameworks should specify when empathetic AI can intervene and when it is necessary for humans to take full control” (379).

Maintain human autonomy and balanced reliance on AI. The authors state, “It is vital to maintain human autonomy, so that AI does not steer our decisions in unwanted directions” (p. 378). They also conclude that “one should neither rely too little nor too much on the individuality of AI decision-making” (p. 376).

Use emotional and behavioral analysis to support human-directed lines of inquiry. For forensic investigations, the authors explain that AI “interprets the emotional and behavioral context surrounding communications or suspicious activities” (Baltezarević et al., 2026, p. 378). With human involvement, it “can recommend lines of inquiry that balance analytical precision with the human context of incidents” (p. 378).

Protect emotional data against misuse. The authors direct that “emotional data must be treated with the same seriousness as, for example, transactional or behavioral data” (Baltezarević et al., 2026, p. 378). They further state, “It must also be ensured that data on emotional reactions is stored and protected from any misuse” (p. 379).

Require understandable explanations of AIʼs analytical and emotional insights. The authors emphasize that “Grasping how models make decisions is crucial” when identifying unusual behavior (p. 377). For empathetic systems, they additionally state that “empathetic AI systems need to be upfront about their emotional insights, not just their threat detection abilities” (p. 379).

Use adaptive assistance to reduce cognitive burden. The article describes how, in “SOCs or forensic teams,” AI can “adjust their communication and support based on the human operatorʼs emotional and cognitive load” (p. 376). Examples of adaptive assistance include “downplay less urgent alerts, offer concise insights, or even recommend collaborative triage” when analysts are overwhelmed (p. 375).

Train examiners to understand AI feedback and emotional cues. The authors state, “Training employees to better understand feedback received through artificial intelligence is key”. They further recommend that “during analyst training, top priority must be given to emotional literacy” (p. 379).

Preserve situational awareness and decision-making skills. The article makes the usefulness of empathetic AI conditional on its not affecting analystsʼ “individual situational awareness” and not undermining “their decision-making skills” (p. 380). These conditions should guide examinersʼ continued use of AI assistance.

Evaluate effectiveness and long-term effects. The authors recommend creating “metrics to measure the effectiveness of emotional AI” and developing “standardized frameworks in security operations for managing emotional data” (p. 379). They also identify the need to understand effects on “human competence, self-confidence, short-term productivity gains and, perhaps most importantly, cognitive resilience” (p. 380).

Issues, Challenges and Pitfalls

Incorrect interpretation of emotional cues. The authors identify “making sure that emotional cues are understood correctly” as a principal concern and warn against “false positives or negatives” (p. 378). Misinterpretation “can produce incorrect suggestions that, in addition to undermining trust, negatively affect efficiency” (p. 379).

Privacy risks and inadequate safeguards. The article states, “Whenever AI processes sensitive data for security reasons, privacy concerns arise”. It also cautions that “safeguards for the use of this private information arenʼt always the best ones” (p. 378).

Deliberately misleading emotional signals. The authors warn that “malicious actors can provide misleading emotional signals, which can compromise the monitoring of the system” (Baltezarević et al., 2026, p. 378). They explain that false cues could “bypass security measures and influence distorted decision-making” (p. 379).

AI-enabled social engineering and deepfake deception. The article describes how fraudsters can “exploit technology that mimics human empathy to break trust and bypass defense systems” (p. 379). It also identifies “deepfake attacks that alter audio and visual data” among AI-enabled threats (p. 374).

Excessive trust in apparently definitive responses, The authors caution that “decision-making and user trust can be largely disproportionate to the apparent definitiveness of AI responses” (p. 376). They additionally observe that “AI cues can easily deceive people into believing theyʼre interacting with sentient beings” (p. 378).

Overreliance and deterioration of examiner skills. The authors report that analystsʼ increasing reliance on AI decisions “can cause a risk of overburdening and the reduction and dulling of skills” (p. 380). They also warn that “over-reliance on it or misinterpretation of emotional cues can have the opposite effect” from the intended benefits (p. 379).

Empathic error and alienating interactions. The article cautions that “even highly skilled demonstrations of empathy have the potential to alienate recipients, and there is always the possibility of empathic error” (p. 374). This is an identified limitation of using empathetic interaction as an investigative support mechanism.

Dependence on human expertise and available information. For the human-in-the-loop learning approach, the authors explain that “the effectiveness of this learning process largely depends on the expertise of the human team and the availability of pertinent information” (p. 376).

Limited practical evidence and unresolved long-term effects. The authors identify the need for further research into “the practical consequences, reliability and moral implications of these systems” (p. 379). Their conclusion emphasizes that the field “is still in its early stages of development” (p. 380).

AI Use Statement

Perplexity AI was used in the development of this information.

References

Baltezarević, R., Stanojević, M., Izgarević, D., & Azdejković, M. (2026). Enhancing human machine cooperation through the integration of empathetic AI into corporate cybersecurity and forensic operations. International Journal of Cognitive Research in Science, Engineering and Education (IJCRSEE), 14(2), 373–384. https://doi.org/10.23947/2334-8496-2026-14-2-373-384

Perplexity [Generative AI tool]. https://www.perplexity.ai/

Cite this page:

Kardasz, F. (2026, October). AI & Digital Forensics - Review of Baltezarević et. al. (2026) - Enhancing Human-Machine Cooperation .....https://kardasz.blogspot.com/2026/10/ai-digital-forensics-review-of.html

Sunday, October 11, 2026