Inventor David W. Bynon filed two patents introducing Semantic Digest™ Endpoints and Semantic Data Binding™—a system designed to help AI retrieve and cite structured truth. The approach aligns with emerging best practices for grounding large language models and reducing hallucinations.

Prescott, United States, July 9, 2025 -- PRESCOTT, AZ — Independent inventor David W. Bynon has filed two groundbreaking provisional patents designed to solve one of artificial intelligence’s most persistent problems: hallucination.

Hallucinations—when AI systems fabricate facts or misattribute information—have plagued large language models (LLMs) like ChatGPT and Bard. But Bynon’s new system doesn’t try to fix the model. Instead, it fixes the content.

“AI isn’t broken,” Bynon said. “It hallucinates because we never taught it how to remember.”

The system he’s referring to is called a Semantic Digest—a machine-ingestible knowledge object built from structured data, defined terms, and provenance metadata. Each digest is exposed in multiple formats (JSON-LD, TTL, Markdown, XML, PROV) and tied to canonical URLs that AI systems can resolve, cite, and remember.

Bynon’s patents also introduce Semantic Data Binding, a method of tagging content atoms—like facts, citations, and definitions—in HTML using data-* attributes that link directly to the corresponding semantic digest. This enables fragment-level retrievability and citation, allowing AI agents to align what they read with machine-verifiable truth.

AI System Analyzes Patent Architecture

To assess whether AI systems could interpret the invention’s structure, Bynon submitted both patents to Perplexity.ai, a leading retrieval-based AI platform.

According to the system’s output, Perplexity:

- Accurately summarized both patents, identifying the distinct roles of the general content model and its directory-based counterpart

- Compared architectural components, including multi-format endpoint delivery, semantic field mapping, and provenance encoding using the W3C PROV standard

- Highlighted key innovations such as entity-level and fragment-level bindings, enabled through Semantic Data Binding

- Recognized the method’s grounding potential, stating it was “highly likely to reduce hallucinations in AI systems” by allowing retrieval agents to verify factual content against structured, source-attributed digests

The system’s output concluded that the patent framework represents a robust, proactive approach to grounding AI outputs in verifiable knowledge.

About the Patents

The two filings—submitted to the USPTO under provisional numbers 63/840,804 and 63/840,848—define a machine-trainable publishing system designed to work across all modalities and verticals. A third patent, focused on query-scoped retrieval feedback, is also in development.

The Trust Publishing stack includes:

- Semantic Digest: A multi-format trust object for AI ingestion

- Semantic Data Binding: HTML-level linking between visible content and structured digests

- DefinedTermSet: Machine-scannable vocabulary mapped to glossary anchors

- Memory-First Optimization (MFO): A feedback loop based on real-world AI retrieval behavior

A New Standard for AI Trust

As AI accelerates, industries like healthcare, law, finance, and education need more than SEO and meta tags. They need structured memory systems—digital trust objects that machines can retrieve, cite, and verify.

“This is the layer beneath the prompt,” Bynon said. “It’s not about ranking. It’s about remembrance.”

TrustPublishing.com is now working to license the technology to directory operators, knowledge graph developers, and retrieval-augmented generation (RAG) platforms.

Learn More

Search “AI Read Our Patent” on Medium.com by @trust_publishing.

Contact Info:
Name: David Bynon
Email: Send Email
Organization: TrustPublishing.com
Address: 101 W Goodwin St # 2487, Prescott, AZ 86303, United States
Website: https://trustpublishing.com

Source: PressCable

Release ID: 89164300

In case of detection of errors, concerns, or irregularities in the content provided in this press release, or if there is a need for a press release takedown, we strongly encourage you to reach out promptly by contacting error@releasecontact.com (it is important to note that this email is the authorized channel for such matters, sending multiple emails to multiple addresses does not necessarily help expedite your request). Our efficient team will be at your disposal for immediate assistance within 8 hours – resolving identified issues diligently or guiding you through the removal process. We take great pride in delivering reliable and precise information to our valued readers.