Most Recent Additions*

PDF

AI Safety for Agentic AI Tool Use: Preventing Prompt-Injected Tool Calls and Data Purpose Laundering with Hardware-Verified, Per-Action Execution Control for MCP, Computer-Use, Coding Agents, Memory Writes and Cloud-Assisted AI — Technical Relevance to Claude, ChatGPT, Gemini, Copilot, Apple Intelligence and Meta AI
Sangam Kumar Das

PDF

AI Safety for On-Device AI Agents: Privacy-Preserving Assistant Interoperability with Per-Action, Hardware-Verified Authorization for Messages, Payments, File Exports and Sensor Data — Technical Relevance to Apple Siri, Google Gemini, Microsoft Copilot, Samsung Galaxy AI and Meta AI
Sangam Kumar Das

PDF

AI Safety for AI Agents: Hallucination-Proof, Non-Completable Execution Control for Tool Calls, Payments and Real-World Actions — Technical Relevance to Anthropic Claude, OpenAI ChatGPT, Google Gemini, Microsoft Copilot and Amazon Bedrock Agents
Sangam Kumar Das

PDF

Cross-Platform Agent Mailboxes and Scheduled MCP Meeting Sessions
Joe Boll

PDF

Decoy API Responses: Counterattacks, Poisoned Outputs, and Malicious Honeypots, with Safeguards for Good-Faith Users
Joe Boll

PDF

Basin Water Record: A Shared, Signed Record per Water Basin in which Large Users' Certified Meter Readings Are Cross-Referenced with the Utility's Declared Supply and a Published Order of Curtailment, a Declared Shortage Automatically Tightens Each Committed User's Limit Without Waiting for an Order, Actual Cuts Are Logged and Checked Against the Published Order so that Any Inversion of Priority Is Visible, and Desalinated Supply Is Recorded with Its Energy and Brine and Never Counted as an Offset
Edward D. H. Wang

PDF

Low-Bandwidth Semantic Video Transfer with Local Generative Reconstruction
Joe Boll

PDF

The Paradox of the Mass Deficit in Nuclear Fusion
Peter Stevenson

PDF

The Abyssal Bioluminescence Coherence Paradox
Peter Stevenson

PDF

Technical Whitepaper - Secure and Decentralized Retrofit for Industrial IoT Devices
Tiziano Ciccone

*Updated as of 11/10/26.