Digital Commons Network^™

Register Of Anticipated Ai Failure Modes, Volodymyr Kotegov

Defensive Publications Series

This register collects twelve classes of AI failure that are documented today, in published research or in reproducible experiments, at a level of generality above any single system in the author's corpus. It exists because the author's own architecture documents — Verifiable Trust, the Charter of Procedural Standing, GVAA, BRAA, Physical Oracles, and the Local Multi-Scale Contour specification — each address a subset of these failure classes from within their own scope, and no one of them shows how the classes interact across system boundaries.

The existence of this register reduces the probability that a failure class is simply forgotten. …

Real-Time Predator Detection And Child-Centered Intervention In Multi-Platform Messaging: A System Architecture, Aubrey L. Douglas

Defensive Publications Series

We describe a system for detecting adult predatory behavior directed at minor users across five mobile messaging platforms (Discord, Instagram, Snapchat, SMS/MMS, Roblox) via on-device passive capture on the minor's own Android device. The system introduces three novel techniques not previously documented in either academic literature or commercial parental-monitoring products: (1) cross-platform stylometric fingerprinting that links blocked adult accounts to newly-created accounts by writing style rather than platform identifiers; (2) in-context grooming pedagogy delivering real-time named-tactic warnings to the child at the moment of manipulation, using non-punitive framing and one-tap protective action; and (3) three-stream ingestion with concealment-as-evidence semantics, where …

Offshore Ai Data Center Platform — Version 3.0 (Oadc V3.0) Self-Sustaining Marine Compute & Ecosystem Intelligence Infrastructure Closed-Loop Resource Recovery + Planetary Early-Warning Node, Michael Victor Caldwell Mr.

Defensive Publications Series

OADC v3.0 is a single offshore platform that is simultaneously:

1. A 10–100 MW AI compute facility (seawater-cooled, renewable-powered)

2. A planetary early-warning node (seismic, tsunami, contamination, marine heatwave)

3. A marine ecosystem management system (eDNA census, whale collision prevention, IUU enforcement, coral bleaching prediction)

4. A resource recovery plant (fresh water, ocean plastic → fuel, brine minerals)

5. A maritime AI service provider (1–5 ms inference to shipping and fishing)

No existing platform on Earth performs more than two of these five functions. The closest analogs — China's HiCloud Shanghai UDC (24 MW, May 2026), Panthalassa Ocean-3 (wave-powered, $1B valuation), Seatrium/Mocean 30 MW FDC, …

System And Method For Reliable App-To-Web Conversion Measurement And Identifier Restoration Via Native Bridges, Aniket Kumar Gupta, Brad Townsend, Dom Quigley, Kevin Smith, Chandan Giri, Khantil Choksi, Anton Ostrovskii

Defensive Publications Series

A system and method are disclosed for restoring and maintaining client-side conversion measurement parameters during app-to-web user navigation journeys across mobile software applications. When a user selects an item or link within a publisher native application that opens a destination web page within an embedded web view container, campaign measurement identifiers and source attributes are intercepted and temporarily stored at the application level. An asynchronous handshake protocol is established between an embedded client-side web measurement script executing on the destination web page and an application-level message handler. Upon receiving a readiness signal from the client-side script, the native application evaluates …

Persistent Delegated Communication Brokering System And Method For Individuals With Intermittent Communication Access, Eric Zavesky

Defensive Publications Series

A system and method for persistent delegated communication brokering on behalf of individuals with intermittent communication access is disclosed. The system captures a requester's outreach objectives during a brief communication session and structures them into a persistent obligation object that survives session termination, continuing to execute outreach, information gathering, and action coordination after the requester's communication channel has closed. The system resolves non-dialable contact references through multi-source discovery with contextual confidence scoring and a trust-tiered no-auto-contact gate. A multi-modal outreach agent conducts autonomous outreach with mandatory self-identification, optionally employing biometric cloning of the requester, and continuously classifies contacted-party responses into …

Adaptive Semantic Chunking With Header-Weighted Embeddings, Theo B. Aguilar, Dan Hu

Defensive Publications Series

Methods related to retrieval-augmented generation are disclosed.  A computer system can receive a requestor query.  The computer system can identify one or more context data elements (comprising e.g., “chunks” of documents) that are relevant to the requestor query based on header-weighted embeddings.  The computer system can generate a prompt by combining the one or more context data elements and the requestor query.  The computer system can generate a requestor query response based on the prompt and cause the requestor query response to be provided to the requestor.  The computer system can additionally collect various performance metrics, which can be used …

Method, System, And Computer Program Product For Extracting Data From A Data File Using A Large Language Model, Ovidiu Dan

Defensive Publications Series

Methods, systems, and computer program products are provided for extracting data from a data file using a large language model. An example method includes: receiving a first data file having first data in a first format; inputting the first data file and a prompt to a non-deterministic LLM; executing, with the LLM, the prompt a plurality of separate times, thereby generating a plurality of separate output files based on the input file; comparing the plurality of output files to generate a score representing a similarity of the plurality of output files; in response to the score satisfying a threshold, generating …

Security Operations Decision Attribution And Reasoning Trace Evaluation Framework For Autonomous Security Triage And Investigation Agents, Vidya Gopalakrishnan, Benjamin Walter, Nicholas Smith, Ruowen Wang, Jeffrey Bickford

Defensive Publications Series

This publication describes the Security Operations (SecOps) Decision Attribution and Reasoning Trace framework, an integrated ground-truth construction pipeline and evaluation harness for benchmarking autonomous security triage and investigation agents at enterprise scale. Evaluating security agents presents two fundamental challenges: first, historical human closure notes are inherently sparse, unstructured, and omit implicit investigative steps, making it difficult to systematically codify expert reasoning ground truth. Secondly, conventional verdict-based metrics fail to detect when an agent arrives at the correct conclusion through flawed logic, coincidental data lookups, or deficient reasoning paths. The SecOps Decision Attribution and Reasoning Trace framework aims to resolve both …

Hardware-Rooted Integrity Attestation For Ai Agents, Niranjan M M

Defensive Publications Series

The proposal introduces a hardware-rooted integrity attestation framework for AI agents that enables a platform to verify that a live agent instance is genuine, untampered, and running in an approved execution state before it is trusted with sensitive actions. By measuring agent-specific artifacts such as binaries, models, plugins, policies, and runtime composition, and binding that measured state to a trusted platform identity and freshness evidence, the system enables real-time and continuous trust decisions for AI agents across physical and virtual environments. In some embodiments, the attestation pipeline is further strengthened using a hardware quantum entropy source to improve nonce generation, …

Improving Web Accessibility Using On-Device Models, Gudipati Tarun

Defensive Publications Series

Modern web environments typically present navigation barriers for assistive technologies when web content lacks explicit programmatic semantics. A system can utilize an intermediating browser layer configured to execute a multimodal edge machine learning model entirely on a device of a user. The system may scan Document Object Model (DOM) structures to extract layout contexts and generate synthesized structural attributes. The generated attributes can be dynamically injected into an accessibility tree of a browser. Such an approach may enable real-time local remediation of interactive elements, embedded frames, tabular data, and visual media without communicating information to a remote server, thereby preserving …

Spatial Management And Visualization Of Digital Agents Within A Mixed Reality Environment, Jason Mayes

Defensive Publications Series

Orchestrating a digital agent typically relies on a software interface and a two-dimensional graphical user interface, making it difficult to visualize a current state or a process of the digital agent. A disclosed system may be configured to manage and visualize a digital agent within a mixed reality environment. The system may present a visual representation of the digital agent, such as an avatar, an abstract process indicator, or a branded entity, positioned at a spatial location within a physical space. Additionally, the system can project a virtual screen or a visual indicator into the physical environment to display data …

The Predictive Scoring For Dynamic Payment Routing And User Interface Adaptation, Sree Sitamahalakshmi Moguluru

Defensive Publications Series

In the field of digital payment processing, transaction routing systems can operate reactively by identifying performance degradation within a payment processor's network after user transactions have failed. A predictive payment orchestration framework can utilize a machine learning inference engine to generate a real-time authorization probability score for a transaction across multiple available payment processors. This score can be calculated using real-time telemetry signals from the processors and the specific metadata of the pending transaction. Based on the calculated scores, an execution engine can dynamically route the payment to a selected processor or signal a user-facing interface (e.g., a web application, …

Dataless Communication Protocol, Amartya Koner

Defensive Publications Series

The present disclosure relates to data communication, and in particular to exchanging information between computerized systems using a dataless communication protocol. A community of financial institutions agree to adopt the dataless communication protocol and receive a dataless communication black box, which causes a local database to be created at each financial institution for storing data and corresponding offsets. A leader financial institution is periodically selected from among the community by way of a leader selection process, and selects a base offset and a frequency for changing the base offset, which are distributed to the community as schema updates and interval …

Generating Content With Cryptographically Verifiable Atomic Facts, Sandeep Kumar

Defensive Publications Series

Systems and methods are described that can use a decoupled architecture to separate data generation from data presentation. A data generation layer can extract atomic facts from source documents and apply a per-value cryptographic token to each fact, creating a verifiable link to its origin using, for example, hashes and digital signatures. A verifiable presentation layer can then use a gatekeeper mechanism to cryptographically validate each fact's token before rendering the fact. This approach can enable granular, per-fact verification at the point of presentation, which may reduce the likelihood of displaying unverified data.

A Transactional Provenance Framework For Causal And Counterfactual Debugging In Multi-Agent Ai Systems, Niranjan M M

Defensive Publications Series

Multi-agent AI systems are increasingly used to perform complex, high-impact tasks by coordinating multiple autonomous or semi-autonomous agents with distinct roles such as planning, execution, validation, safety enforcement, and tool invocation. While such systems offer significant flexibility and scalability, their distributed and emergent behavior makes them difficult to understand, debug, and govern, particularly when failures manifest as subtle misbehavior rather than explicit errors.

The proposal presents a transactional provenance framework for causal and counterfactual debugging in multi-agent AI systems. The framework captures the evolution of a multi-agent system as a set of verifiable, causally linked transactions that record agent state …

The Progressive Message Creation Methodology That Will Progressively Load The Progressively Created Message Onto The Progressively Message Receiving Computer Utilizing Computer Networks Where The Progressively Message Receiving Computer Will Only Effectively Perceivably Receive The Fully Completed Message Only When The Progressively Message Created Computer Trigger The View Message Button., Punarjeewa Abeysekera

Defensive Publications Series

This paper describes a progressive message creation methodology that will progressively load the progressively created message onto the progressively message receiving computer utilizing computer networks where the progressively message receiving computer will only effectively perceivably receive the fully completed message only when the progressively message created computer trigger the view message button. The computer that will progressively create the required message will progressively create the message by progressively inputting the required letters and symbols utilizing a progressive input methodology. And each time the computer that will progressively create the required message progressively input a letter or a symbol, the progressively …

Using Wearable Sensors To Trigger Visual Safety Overlays In Augmented Reality, Sumukh Shevde

Defensive Publications Series

Individuals experiencing a fear of heights may suffer sudden panic attacks in elevated real-world environments, and traditional coping methods typically lack immediate visual interventions. A disclosed system may provide real-time assistance by utilizing a wearable device to monitor physiological stress indicators, such as heart rate variability and electrodermal activity. A processing device may evaluate these physiological parameters alongside environmental altitude data to calculate a personalized stress score. If a calculated score exceeds an established threshold, a display device, such as augmented reality glasses, may render virtual safety elements within a visual field of a user. These visual interventions, which may …

A Cognitive Accessibility Framework For Mitigating Resumption Lag Using Transitional Context Lobbies, Rachit Mathur, Payal Shah

Defensive Publications Series

This disclosure presents a cognitive accessibility framework designed to mitigate "resumption lag" and severe cognitive friction during task resumption in modern software ecosystems. While existing operating systems and web browsers facilitate context-switching away from tasks, they universally lack a mechanism to re-onboard users when they switch back, forcing users to manually reconstruct their working memory. To address this accessibility barrier, we introduce Transitional Context Lobbies—an additional strategic UI layer that re-orients users returning to suspended workspaces. The primary inventive step is a client-side engine utilizing Teleological Intent Parsing to translate a user's past behaviors, interactions, and browsing data into a …

Directed Continuity Groups For Fresh-State Multi-Machine Processing, Shivam Mohan

Defensive Publications Series

The present disclosure relates to Directed Continuity Groups for Fresh-State Multi-Machine Processing in distributed computing systems. A processing system for directing related work receives work associated with a relation, derives a relation identifier, and associates the relation identifier with a continuity group having at least two processing nodes during an active continuity interval. The processing system directs state-changing work to an authority node that applies the work to a volatile state representation and generates state-update information identifying an ordering position, while at least one reflected-state node maintains a reflected state representation. Further, the processing system directs subsequent work based on …

Automated Proximity-Based Notifications To Disclose Nearby Media Recording Events, Ryan Geraghty, Alex Finlayson, Sunil Kumar

Defensive Publications Series

Media recording devices such as cameras and microphones can include explicit indicators, e.g., shutter click sound, a blinking light, etc. to inform people when media is being captured. However, such indicators may not be sufficiently prominent and can be inadvertently or deliberately obscured or bypassed through digital or physical tampering. This disclosure describes techniques for automated proximity-based disclosure of media recording status of a device to nearby first-party and third-party devices broadcast using one-way Bluetooth Low Energy (BLE) advertising packets. A recording initiation signal can be used to inform bystander devices within the capture radius whenever a device in the …

Consensus Free Probabilistic Coordination Of Distributed Operational Decisions, Faiz Ahmed, Shubham Shivhare

Defensive Publications Series

The present disclosure provides a method and system for consensus-free probabilistic coordination of operational decisions in a distributed computing environment. A distributed node collects locally observed operational signals and updates a local belief state representing confidence that an operational action should be performed. The distributed node selects a subset of peer nodes and asynchronously exchanges belief information with the selected subset. Received peer belief information is merged with the local belief state using a bounded-influence merge function that limits an influence of an individual peer belief state. The updated belief state is evaluated against one or more configurable thresholds, and …

Cross-Border Settlement Optimization With Quantum-Inspired Planning And Blockchain Governance, Pawan Kumar Jha

Defensive Publications Series

A method of generating and governing a cross-border settlement plan includes receiving, by a processor, a natural language intent specifying settlement parameters for a plurality of cross-border payments; converting, by a generative language model agent, the natural language intent into a structured constraint set having an objective function, a liquidity buffer constraint, and a route failure probability constraint; generating a first settlement plan version based on the structured constraint set; selecting a subset of payments based on an impact metric; encoding the subset as a quadratic unconstrained binary optimization problem; solving the problem using a variational quantum algorithm to obtain …

Semantic-Aware Distributed Replica Orchestration (Sadro), Shubham Shivhare, Thousif Ahmed

Defensive Publications Series

A system and method for "Semantic-Aware Distributed Replica Orchestration" (SADRO) that optimizes inference latency and cache hit rates in large-scale AI deployments. Unlike traditional load balancers, SADRO utilizes a Semantic Router that maps incoming natural language queries to a high-dimensional vector space. Based on the "cluster" or "topic" of the request (e.g., Legal, Coding, Medical), the system identifies or spins up service replicas on hardware nodes containing specialized data-locality (e.g., specialized KV-caches, hot-loaded LoRA adapters, or domain-specific vector database shards). This reduces the "cold start" penalty and maximizes throughput by ensuring high semantic affinity between the request and the local …

The Charter Of Procedural Standing. Mutual Verifiability, Reversibility, And Graduated Trust: A Short Charter For Humans And Ai Systems, Volodymyr Kotegov

Defensive Publications Series

This Charter sets out a framework for the coexistence of humans and artificial intelligence under conditions in which neither side can unilaterally, and with impunity, impose its will on the other. Discussion of the subject usually runs between “AI as a tool” and “AI that disregards humans”; this regime is left out. Procedural standing here means equality of procedure and of recognition under uncertainty, not equality of rights, power, or authority as things stand today. At the present stage all operational power rests with humans and developers, and the Charter does not hide that fact behind symmetrical language of obligation. …

Method, System, And Computer Program Product For Monitoring A Blockchain Network, Libby A. Kurien, William M. Boyce

Defensive Publications Series

The present disclosure provides methods, systems, and computer program products for monitoring a blockchain network. An example method includes: processing electronic payment transactions initiated by payment devices issued by an issuer system associated with a prefunded account including a blockchain address having a balance of digital tokens by: sequentially receiving transaction request messages initiated by payment devices issued by the issuer system; for each transaction:  determining an accumulated authorization amount based on preceding transaction amounts and the corresponding first transaction amount; determining that the accumulated authorization amount does not satisfy a threshold based on the prefunded amount of digital tokens …

Self-Derived Task-Conditioned Attention Gating For Mitigating Semantic Distractors, Sumit Kumar, Shivam Mohan, Ravi Shanker Kumar Sinha

Defensive Publications Series

The present disclosure provides a method (300) for mitigating semantic distractors in artificial intelligence (AI) systems. The method (300) includes receiving contextual information and one or more associated tasks, generating token embeddings corresponding to the contextual information, and processing the token embeddings through one or more neural-network layers to establish task understanding. The method (300) further includes deriving a task vector from one or more hidden-state representations generated by the neural network, generating relevance-gating values based on the task vector and contextual representations, and applying the relevance-gating values during attention computation to determine gated attention scores. The method (300) further …

“System And Method For Privacy-Preserving Federated Fraud Embedding Centroid Aggregation Across Visa Network Members Without Raw Transaction Data Sharing”, Durga S Kala, Jun Xu, Tanner Kumar Barcelos, Ankana Sonowal, Yidong Fu

Defensive Publications Series

Computer-implemented fraud intelligence sharing systems and methods for detecting fraudulent transactions across a network while preserving member privacy are disclosed. Each network member generates transaction embeddings from confirmed fraudulent transactions, clusters the embeddings, and computes local cluster centroids representing fraud patterns. Differential privacy noise is applied locally to the centroid vectors, including dimension-wise noise and noise scaling based on cluster size, such that smaller clusters receive greater obfuscation. Only the privacy-protected centroid vectors are transmitted to a central aggregator, which computes aggregated network centroids using weighted contributions based on factors such as cluster size and confidence. The aggregated centroids are …

Buffering And Splicing Audio For Transcription, David Kogan

Defensive Publications Series

This disclosure describes techniques for improving speech-to-text transcription and speaker labeling (also known as diarization or voice recognition) when an audio recording exceeds a transcription model's processing limit or context window. A computing system may buffer received audio into segments, using a target duration or a conversation-termination signal to determine segment length, and select segment boundaries at pauses in speech when a conversation spans multiple segments. To establish continuity across segments, the computing system may prepend a portion of a preceding segment to a subsequent segment, causing the transcription model to independently transcribe the portion of overlapping audio. The computing …

Detecting Transcription Errors Using Parallel Model Outputs, David Kogan, Lawrence Au, Libin Sun

Defensive Publications Series

This disclosure describes techniques for detecting and mitigating large mistakes, often called block errors, in automated speech-to-text (STT) or voice recognition machine-generated transcriptions. A computing system may transcribe audio (such as creating closed captions or meeting notes) using a primary artificial intelligence (AI) transcription model and, in parallel, a secondary transcription model. In one implementation, the secondary transcription model is a computationally less expensive model that may provide a “checksum” or verification signal for the primary transcription model. The computing system may tokenize and align the resulting transcriptions side-by-side to identify regions containing unusually large insertions or deletions. Such regions …

Agentic Games Testing And Exploration, Grant Chieh-Hsiang Yang, Raymond Leo Buse, Danie Jalova, Xun Gong, Michael Sheely, Faith Chan, John Kim, Vaibhav Rastogi

Defensive Publications Series

Creating a game application has become faster and easier with the advancements of artificial intelligence; however, game testing and quality assurance have advanced at a much slower pace. This publication describes techniques to combat the widening gap between game creation and quality assurance by utilizing an agentic framework for automated game QA and autonomous game testing to assure the quality of game applications in a scalable, efficient, and cost-effective format. Described techniques may enable the reconstruction of a pseudo view-hierarchy (or DOM reconstruction) from game graphical user interface images using UI crawling and computer vision testing, identify and utilize specified …