Intellectual property
Patents
One issued U.S. patent. Two pending applications. The technical foundation for human-aligned AI safety.
US10,380,912 B2 — AI-Powered Language Learning System
Issued · Active through 2037
An AI-powered language learning system that lets users create their own personalized learning content using voice recognition, image recognition, and automated content generation. The invention mimics how humans naturally acquire language — through immersive, contextual, user-generated exposure — rather than through rote vocabulary drills. Sole inventor and assignee: Mongkol Thitithamasak.
USPTO Application No. 64/078,432 — Biometric-Authenticated Post-Quantum Identity Verification for AI
Patent pending · Filed 29 May 2026 · Attorney Docket MONGKOL-HASF-001
The first of two pending applications covering the Humanity Alignment Safety Framework (HASF). The invention combines a wearable multi-modal biometric sensor — measuring heart rate, skin temperature, galvanic skin response, motion, and skin contact simultaneously — with post-quantum cryptography to verify human identity at the moment of every AI interaction. Once verified, that identity unlocks personalized AI output across three independent layers: the prompt sent to the model, the generation parameters used, and the post-processing applied before the response is returned. The result is AI that responds differently to a verified human than to an unverified one, with cryptographic guarantees of who is actually present.
Built on NIST-standardized post-quantum algorithms (FIPS 203, 204, and 205), the system is designed to remain secure even against future quantum computers, and supports algorithm upgrades without hardware replacement.
Application not yet published. Public USPTO record will be available 18 months after the filing date.
Attorney Docket MONGKOL-HASF-002 — Safety-Constrained Control of Embodied Robotic AI & Participatory Alignment Governance
Patent pending · Attorney Docket MONGKOL-HASF-002
The second HASF application extends the verified-identity framework into two new domains: the physical safety of embodied AI systems, and the governance of AI alignment itself. For robots, autonomous vehicles, and surgical assistants, every behavioral command must carry a fresh, cryptographically-signed authorization tied to a biometrically-verified human — preventing remote command injection and identity spoofing. Each verified user has a personalized safety envelope specifying proximity tolerances, velocity limits, force thresholds, and autonomy boundaries; on any authentication failure the robot enters a safe state that can only be exited with a fresh biometric authentication.
In parallel, the same verification infrastructure powers a Humanity Alignment Platform — a Sybil-resistant mechanism for capturing structured human values, modeled statistically across populations, and translated into multi-dimensional AI Alignment Scores. A single trust score governs both the user’s influence in alignment governance and the maximum autonomy level any robot operating under their authority may exercise. The two functions, normally separated, are unified here as a deliberate architectural choice.
Application not yet published. Public USPTO record will be available 18 months after the filing date.
Licensing & inquiries
The HASF patents are filed to document a coherent technical architecture for trustworthy AI deployment — not to restrict the field. Licensing discussions are welcome from companies building AI systems, robotics platforms, identity infrastructure, or governance tools that would benefit from these primitives.
Contact: licensing@humanaisafety.com
Patent Pending — USPTO Application Nos. 64/078,432 (MONGKOL-HASF-001) and MONGKOL-HASF-002. U.S. Patent No. 10,380,912 issued. This page is provided for general reference and does not constitute legal advice or an exhaustive description of patent claims. Refer to the official USPTO records for legally controlling text.