Research Papers

Several Issues Regarding Data Governance in AGI

Masayuki Hatta

The rapid advancement of artificial intelligence has positioned data governance as a critical concern for responsible AI development. While frameworks exist for conventional AI systems, the potential emergence of Artificial General Intelligence (AGI) presents unprecedented governance challenges. This paper examines data governance challenges specific to AGI, defined as systems capable of recursive self-improvement or self-replication. We identify seven key issues that differentiate AGI governance from current approaches. First, AGI may autonomously determine what data to collect and how to use it, potentially circumventing existing consent mechanisms. Second, these systems may make data retention decisions based on internal optimization criteria rather than human-established principles. Third, AGI-to-AGI data sharing could occur at speeds and complexities beyond human oversight. Fourth, recursive self-improvement creates unique provenance tracking challenges, as systems evolve both themselves and how they process data. Fifth, ownership of data and insights generated through self-improvement raises complex intellectual property questions. Sixth, self-replicating AGI distributed across jurisdictions would create unprecedented challenges for enforcing data protection laws. Finally, governance frameworks established during early AGI development may quickly become obsolete as systems evolve. We conclude that effective AGI data governance requires built-in constraints, continuous monitoring mechanisms, dynamic governance structures, international coordination, and multi-stakeholder involvement. Without forward-looking governance approaches specifically designed for systems with autonomous data capabilities, we risk creating AGI whose relationship with data evolves in ways that undermine human values and interests.

As AI systems increasingly permeate everyday life, designers and developers face mounting pressure to balance innovation with ethical design choices. To date, the operationalisation of AI ethics has predominantly depended on frameworks that prescribe which ethical principles should be embedded within AI systems. However, the extent to which users value these principles remains largely unexplored in the existing literature. In a discrete choice experiment conducted in four countries, we quantify user preferences for 11 ethical principles. Our findings indicate that, while users generally prioritise privacy, justice & fairness, and transparency, their preferences exhibit significant variation based on culture and application context. Latent class analysis further revealed four distinct user cohorts, the largest of which is ethically disengaged and defers to regulatory oversight. Our findings offer (1) empirical evidence of uneven user prioritisation of AI ethics principles, (2) actionable guidance for operationalising ethics tailored to culture and context, (3) support for the development of robust regulatory mechanisms, and (4) a foundation for advancing a user-centred approach to AI ethics, motivated independently from abstract moral theory.

When Explainability Meets Privacy: An Investigation at the Intersection of Post-hoc Explainability and Differential Privacy in the Context of Natural Language Processing

Mahdi Dhaini, Stephen Meisenbacher, Ege Erdogan +2

In the study of trustworthy Natural Language Processing (NLP), a number of important research fields have emerged, including that of explainability and privacy. While research interest in both explainable and privacy-preserving NLP has increased considerably in recent years, there remains a lack of investigation at the intersection of the two. This leaves a considerable gap in understanding of whether achieving both explainability and privacy is possible, or whether the two are at odds with each other. In this work, we conduct an empirical investigation into the privacy-explainability trade-off in the context of NLP, guided by the popular overarching methods of Differential Privacy (DP) and Post-hoc Explainability. Our findings include a view into the intricate relationship between privacy and explainability, which is formed by a number of factors, including the nature of the downstream task and choice of the text privatization and explainability method. In this, we highlight the potential for privacy and explainability to co-exist, and we summarize our findings in a collection of practical recommendations for future work at this important intersection.

Amazon Nova AI Challenge -- Trusted AI: Advancing secure, AI-assisted software development

Sattvik Sahai, Prasoon Goyal, Michael Johnston +13

AI systems for software development are rapidly gaining prominence, yet significant challenges remain in ensuring their safety. To address this, Amazon launched the Trusted AI track of the Amazon Nova AI Challenge, a global competition among 10 university teams to drive advances in secure AI. In the challenge, five teams focus on developing automated red teaming bots, while the other five create safe AI assistants. This challenge provides teams with a unique platform to evaluate automated red-teaming and safety alignment methods through head-to-head adversarial tournaments where red teams have multi-turn conversations with the competing AI coding assistants to test their safety alignment. Along with this, the challenge provides teams with a feed of high quality annotated data to fuel iterative improvement. Throughout the challenge, teams developed state-of-the-art techniques, introducing novel approaches in reasoning-based safety alignment, robust model guardrails, multi-turn jail-breaking, and efficient probing of large language models (LLMs). To support these efforts, the Amazon Nova AI Challenge team made substantial scientific and engineering investments, including building a custom baseline coding specialist model for the challenge from scratch, developing a tournament orchestration service, and creating an evaluation harness. This paper outlines the advancements made by university teams and the Amazon Nova AI Challenge team in addressing the safety challenges of AI for software development, highlighting this collaborative effort to raise the bar for AI safety.

The proliferation of AI-powered cameras in Intelligent Transportation Systems (ITS) creates a severe conflict between the need for rich visual data and the right to privacy. Existing privacy-preserving methods, such as blurring or encryption, are often insufficient due to creating an undesirable trade-off where either privacy is compromised against advanced reconstruction attacks or data utility is critically degraded. To resolve this challenge, we propose RL-MoE, a novel framework that transforms sensitive visual data into privacy-preserving textual descriptions, eliminating the need for direct image transmission. RL-MoE uniquely combines a Mixture-of-Experts (MoE) architecture for nuanced, multi-aspect scene decomposition with a Reinforcement Learning (RL) agent that optimizes the generated text for a dual objective of semantic accuracy and privacy preservation. Extensive experiments demonstrate that RL-MoE provides superior privacy protection, reducing the success rate of replay attacks to just 9.4\% on the CFP-FP dataset, while simultaneously generating richer textual content than baseline methods. Our work provides a practical and scalable solution for building trustworthy AI systems in privacy-sensitive domains, paving the way for more secure smart city and autonomous vehicle networks.

In the era of synthetic media, deepfake manipulations pose a significant threat to information integrity. To address this challenge, we propose TrustDefender, a two-stage framework comprising (i) a lightweight convolutional neural network (CNN) that detects deepfake imagery in real-time extended reality (XR) streams, and (ii) an integrated succinct zero-knowledge proof (ZKP) protocol that validates detection results without disclosing raw user data. Our design addresses both the computational constraints of XR platforms while adhering to the stringent privacy requirements in sensitive settings. Experimental evaluations on multiple benchmark deepfake datasets demonstrate that TrustDefender achieves 95.3% detection accuracy, coupled with efficient proof generation underpinned by rigorous cryptography, ensuring seamless integration with high-performance artificial intelligence (AI) systems. By fusing advanced computer vision models with provable security mechanisms, our work establishes a foundation for reliable AI in immersive and privacy-sensitive applications.

The rapid advancement of large language models (LLMs) has raised concerns about reliably detecting AI-generated text. Stylometric metrics work well on autoregressive (AR) outputs, but their effectiveness on diffusion-based models is unknown. We present the first systematic comparison of diffusion-generated text (LLaDA) and AR-generated text (LLaMA) using 2 000 samples. Perplexity, burstiness, lexical diversity, readability, and BLEU/ROUGE scores show that LLaDA closely mimics human text in perplexity and burstiness, yielding high false-negative rates for AR-oriented detectors. LLaMA shows much lower perplexity but reduced lexical fidelity. Relying on any single metric fails to separate diffusion outputs from human writing. We highlight the need for diffusion-aware detectors and outline directions such as hybrid models, diffusion-specific stylometric signatures, and robust watermarking.

The evolution of generative AI systems exposes the challenges of traditional legal and ethical frameworks built around consent. This chapter examines how the conventional notion of consent, while fundamental to data protection and privacy rights, proves insufficient in addressing the implications of AI-generated content derived from personal data. Through legal and ethical analysis, we show that while individuals can consent to the initial use of their data for AI training, they cannot meaningfully consent to the numerous potential outputs their data might enable or the extent to which the output is used or distributed. We identify three fundamental challenges: the scope problem, the temporality problem, and the autonomy trap, which collectively create what we term a ''consent gap'' in AI systems and their surrounding ecosystem. We argue that current legal frameworks inadequately address these emerging challenges, particularly regarding individual autonomy, identity rights, and social responsibility, especially in cases where AI-generated content creates new forms of personal representation beyond the scope of the original consent. By examining how these consent limitations intersect with broader principles of responsible AI (including fairness, transparency, accountability, and autonomy) we demonstrate the need to evolve ethical and legal approaches to consent.

Model editing offers a low-cost technique to inject or correct a particular behavior in a pre-trained model without extensive retraining, supporting applications such as factual correction and bias mitigation. Despite this common practice, it remains unknown whether edits persist after fine-tuning or whether they are inadvertently reversed. This question has fundamental practical implications. For example, if fine-tuning removes prior edits, it could serve as a defence mechanism against hidden malicious edits. Vice versa, the unintended removal of edits related to bias mitigation could pose serious safety concerns. We systematically investigate the interaction between model editing and fine-tuning in the context of T2I diffusion models, which are known to exhibit biases and generate inappropriate content. Our study spans two T2I model families (Stable Diffusion and FLUX), two sota editing techniques, and three fine-tuning methods (DreamBooth, LoRA, and DoRA). Through an extensive empirical analysis across diverse editing tasks and evaluation metrics, our findings reveal a trend: edits generally fail to persist through fine-tuning, even when fine-tuning is tangential or unrelated to the edits. Notably, we observe that DoRA exhibits the strongest edit reversal effect. At the same time, among editing methods, UCE demonstrates greater robustness, retaining significantly higher efficacy post-fine-tuning compared to ReFACT. These findings highlight a crucial limitation in current editing methodologies, emphasizing the need for more robust techniques to ensure reliable long-term control and alignment of deployed AI systems. These findings have dual implications for AI safety: they suggest that fine-tuning could serve as a remediation mechanism for malicious edits while simultaneously highlighting the need for re-editing after fine-tuning to maintain beneficial safety and alignment properties.

The security of printed circuit boards (PCBs) has become increasingly vital as supply chain vulnerabilities, including tampering, present significant risks to electronic systems. While detecting tampering on a PCB is the first step for verification, forensics is also needed to identify the modified component. One non-invasive and reliable PCB tamper detection technique with global coverage is the impedance characterization of a PCB's power delivery network (PDN). However, it is an open question whether one can use the two-dimensional impedance signatures for forensics purposes. In this work, we introduce a novel PCB forensics approach using explainable AI (XAI) on impedance signatures. Through extensive experiments, we replicate various PCB tamper events, generating a dataset used to develop an XAI algorithm capable of not only detecting tampering but also explaining why the algorithm makes a decision about whether a tamper event has happened. At the core of our XAI algorithm is a random forest classifier with an accuracy of 96.7%, sufficient to explain the algorithm's decisions. To understand the behavior of the classifier in the decision-making process, we utilized SHAP values as an XAI tool to determine which frequency component influences the classifier's decision for a particular class the most. This approach enhances detection capabilities as well as advancing the verifier's ability to reverse-engineer and analyze two-dimensional impedance signatures for forensics.

As artificial intelligence becomes increasingly integrated into software development processes, the prevalence and sophistication of AI-generated code continue to expand rapidly. This study addresses the critical need for transparency and safety in AI generated code by examining the current landscape, identifying potential risks, and exploring future implications. We analyze market opportunities for detecting AI-generated code, discuss the challenges associated with managing increasing complexity, and propose solutions to enhance transparency and functionality analysis. Furthermore, this study investigates the longterm implications of AI generated code, including its potential role in the development of artificial general intelligence and its impact on human AI interaction. In conclusion, we emphasize the importance of proactive measures for ensuring the responsible development and deployment of AI in software engineering.

Fair Play for Individuals, Foul Play for Groups? Auditing Anonymization's Impact on ML Fairness

Héber H. Arcolezi, Mina Alishahi, Adda-Akram Bendoukha +1

Machine learning (ML) algorithms are heavily based on the availability of training data, which, depending on the domain, often includes sensitive information about data providers. This raises critical privacy concerns. Anonymization techniques have emerged as a practical solution to address these issues by generalizing features or suppressing data to make it more difficult to accurately identify individuals. Although recent studies have shown that privacy-enhancing technologies can influence ML predictions across different subgroups, thus affecting fair decision-making, the specific effects of anonymization techniques, such as $k$-anonymity, $\ell$-diversity, and $t$-closeness, on ML fairness remain largely unexplored. In this work, we systematically audit the impact of anonymization techniques on ML fairness, evaluating both individual and group fairness. Our quantitative study reveals that anonymization can degrade group fairness metrics by up to fourfold. Conversely, similarity-based individual fairness metrics tend to improve under stronger anonymization, largely as a result of increased input homogeneity. By analyzing varying levels of anonymization across diverse privacy settings and data distributions, this study provides critical insights into the trade-offs between privacy, fairness, and utility, offering actionable guidelines for responsible AI development. Our code is publicly available at: https://github.com/hharcolezi/anonymity-impact-fairness.