Large Language Models (LLMs) like ChatGPT are now widely used in writing and reviewing scientific papers. While this trend accelerates publication growth and reduces human workload, it also introduces serious risks. Papers written or reviewed by LLMs may lack real novelty, contain fabricated or biased results, or mislead downstream research that others depend on. Such issues can damage reputations, waste resources, and even endanger lives when flawed studies influence medical or safety-critical systems. This research explores both the offensive and defensive sides of this growing threat. On the attack side, we demonstrate how an author can inject hidden prompts inside a PDF that secretly guide or "jailbreak" LLM reviewers into giving overly positive feedback and biased acceptance. On the defense side, we propose an "inject-and-detect" strategy for editors, where invisible trigger prompts are embedded into papers; if a review repeats or reacts to these triggers, it reveals that the review was generated by an LLM, not a human. This method turns prompt injections from vulnerability into a verification tool. We outline our design, expected model behaviors, and ethical safeguards for deployment. The goal is to expose how fragile today's peer-review process becomes under LLM influence and how editorial awareness can help restore trust in scientific evaluation.

We introduce Model-Bound Latent Exchange (MoBLE), a decoder-binding property in Transformer autoencoders formalized as Zero-Shot Decoder Non-Transferability (ZSDN). In identity tasks using iso-architectural models trained on identical data but differing in seeds, self-decoding achieves more than 0.91 exact match and 0.98 token accuracy, while zero-shot cross-decoding collapses to chance without exact matches. This separation arises without injected secrets or adversarial training, and is corroborated by weight-space distances and attention-divergence diagnostics. We interpret ZSDN as model binding, a latent-based authentication and access-control mechanism, even when the architecture and training recipe are public: encoder's hidden state representation deterministically reveals the plaintext, yet only the correctly keyed decoder reproduces it in zero-shot. We formally define ZSDN, a decoder-binding advantage metric, and outline deployment considerations for secure artificial intelligence (AI) pipelines. Finally, we discuss learnability risks (e.g., adapter alignment) and outline mitigations. MoBLE offers a lightweight, accelerator-friendly approach to secure AI deployment in safety-critical domains, including aviation and cyber-physical systems.

We analyze subliminal transfer in Transformer models, where a teacher embeds hidden traits that can be linearly decoded by a student without degrading main-task performance. Prior work often attributes transferability to global representational similarity, typically quantified with Centered Kernel Alignment (CKA). Using synthetic corpora with disentangled public and private labels, we distill students under matched and independent random initializations. We find that transfer strength hinges on alignment within a trait-discriminative subspace: same-seed students inherit this alignment and show higher leakage {τ\approx} 0.24, whereas different-seed students -- despite global CKA > 0.9 -- exhibit substantially reduced excess accuracy {τ\approx} 0.12 - 0.13. We formalize this with subspace-level CKA diagnostic and residualized probes, showing that leakage tracks alignment within the trait-discriminative subspace rather than global representational similarity. Security controls (projection penalty, adversarial reversal, right-for-the-wrong-reasons regularization) reduce leakage in same-base models without impairing public-task fidelity. These results establish seed-induced uniqueness as a resilience property and argue for subspace-aware diagnostics for secure multi-model deployments.

Federated Learning (FL) enables collaborative model training while preserving data privacy by keeping raw data locally stored on client devices, preventing access from other clients or the central server. However, recent studies reveal that sharing model gradients creates vulnerability to Model Inversion Attacks, particularly Deep Leakage from Gradients (DLG), which reconstructs private training data from shared gradients. While Homomorphic Encryption has been proposed as a promising defense mechanism to protect gradient privacy, fully encrypting all model gradients incurs high computational overhead. Selective encryption approaches aim to balance privacy protection with computational efficiency by encrypting only specific gradient components. However, the existing literature largely overlooks a theoretical exploration of the spectral behavior of encrypted versus unencrypted parameters, relying instead primarily on empirical evaluations. To address this gap, this paper presents a framework for theoretical analysis of the underlying principles of selective encryption as a defense against model inversion attacks. We then provide a comprehensive empirical study that identifies and quantifies the critical factors, such as model complexity, encryption ratios, and exposed gradients, that influence defense effectiveness. Our theoretical framework clarifies the relationship between gradient selection and privacy preservation, while our experimental evaluation demonstrates how these factors shape the robustness of defenses against model inversion attacks. Collectively, these contributions advance the understanding of selective encryption mechanisms and offer principled guidance for designing efficient, scalable, privacy-preserving federated learning systems.

Large Language Models (LLMs) are increasingly deployed in mission-critical systems, facilitating tasks such as satellite operations, command-and-control, military decision support, and cyber defense. Many of these systems are accessed through application programming interfaces (APIs). When such APIs lack robust access controls, they can expose full or top-k logits, creating a significant and often overlooked attack surface. Prior art has mainly focused on reconstructing the output projection layer or distilling surface-level behaviors. However, regenerating a black-box model under tight query constraints remains underexplored. We address that gap by introducing a constrained replication pipeline that transforms partial logit leakage into a functional deployable substitute model clone. Our two-stage approach (i) reconstructs the output projection matrix by collecting top-k logits from under 10k black-box queries via singular value decomposition (SVD) over the logits, then (ii) distills the remaining architecture into compact student models with varying transformer depths, trained on an open source dataset. A 6-layer student recreates 97.6% of the 6-layer teacher model's hidden-state geometry, with only a 7.31% perplexity increase, and a 7.58 Negative Log-Likelihood (NLL). A 4-layer variant achieves 17.1% faster inference and 18.1% parameter reduction with comparable performance. The entire attack completes in under 24 graphics processing unit (GPU) hours and avoids triggering API rate-limit defenses. These results demonstrate how quickly a cost-limited adversary can clone an LLM, underscoring the urgent need for hardened inference APIs and secure on-premise defense deployments.