Authority.inc Launches 12-Month Research Partnership with OxBridge AIx to Redefine Ethics in LLM Search for Procurement

Stockholm, Sweden – December 5, 2025 – Authority Inc. ("Authority.inc") today announced a 12-month research partnership with Oxbridge AIx, a premier AI entrepreneurship community focussed on Oxford Cambridge, focused on truthfulness, fairness, and transparency in LLM-powered vendor search and ranking.

Authority.inc is not only building the authority layer of the internet for B2B procurement, it is also operating as a research lab dedicated to understanding how large language models (LLMs) should interact with verified, structured data when real money and real contracts are on the line.

Through this partnership, Oxbridge AIx will facilitate Authority.inc's collaboration with AI research groups across Oxford and Cambridge, starting with us becoming a Research Sponsor of Oxford Artificial Intelligence Society and sponsoring research with their Safety and Governance Team, to jointly investigate how LLMs behave in procurement workflows and how ranking algorithms can meet rigorous ethical and scientific standards.

Authority.inc as an Applied Research Lab for AI-Mediated Procurement

Authority.inc is built on a simple but demanding thesis:

When AI answers "Who should we buy from?" the result should be as carefully designed and evidence-based as a financial audit, not a casual search result.

To achieve that, Authority.inc operates as an applied research lab at the intersection of:

• LLM behavior and prompt-level retrieval
• Ranking algorithms for B2B vendors
• Truthfulness, bias, and explainability in AI-mediated decisions
• Machine-readable, verifiable vendor facts (JSON-LD, APIs, and catalogs)

The Oxbridge AIx partnership embeds this work into a broader academic context, connecting Authority.inc's live product environment with frontline research at Oxford and Cambridge.

"We think of Authority.inc as part product company, part research lab. Our job is to figure out how LLMs should reason over verified vendor data, not just how they can. Partnering with Oxbridge AIx, and working with Oxford Artificial Intelligence Society's Safety and Governance Team, gives us the scientific counterpart we need: rigorous experiments, critical scrutiny, and peers who are comfortable asking, 'Is this ranking actually truthful and fair?'"

A 12-Month Research Program with Oxbridge AIx

The partnership with Oxbridge AIx is structured as a 12-month research program. Oxbridge AIx will coordinate Authority.inc's work with academic teams, beginning with our role as Research Sponsor of Oxford Artificial Intelligence Society and our sponsorship of research with their Safety and Governance Team, to explore three core areas:

1. Truthfulness in LLM Rankings for Procurement

• Stress-test how LLMs answer complex procurement questions (e.g., "best cyber insurance for large omnichannel retailers with PCI-DSS and SOC2," "top payment providers for Finnish marketplaces with specific bank integrations," etc.).
• Measure hallucinations, missing constraints, and overconfident but incorrect recommendations.
• Formalize what "truthfulness" means when a model recommends vendors based on structured evidence, not just training data.

2. Bias and Fairness Across Vendors

Quantify to what extent LLM answers over-index on:

• Global mega-brands
• English-only sources
• Highly marketed or SEO-optimized vendors

Explore how verified, timestamped vendor data (Authority.inc's graph) can correct or reduce those biases. Develop metrics that distinguish between "popular vendors" and "best-fit vendors for this exact use case".

3. Explainability and Ethical Ranking Protocols

• Design ways for LLM-based rankings to show their work: evidence, constraints, and tradeoffs.
• Propose an ethical ranking protocol for procurement:
  • Minimum evidence rules
  • Citation and recency requirements
  • Transparency about uncertainty and coverage gaps
• Translate this protocol into concrete ranking algorithms and APIs that Authority.inc will deploy in production.

From Research to Product: Building a Trustworthy Authority Layer

This is not a "research in a vacuum" project. The 12-month program is designed to continuously feed back into Authority.inc's live systems, including:

Ranking Engine

Updating Authority.inc's ranking logic for vendor lists, comparison pages, and embeddings-based search to reflect the research-backed protocol for truthfulness and fairness.

LLM-Ready Schemas and APIs

Refining Authority's JSON-LD and API layer so that LLMs and AI agents can:

• Query verifiable vendor facts
• Receive ranking outputs that are explainable, scored, and evidence-linked

Ethics & Review Playbooks

Creating internal checklists and review processes for:

• Shipping new ranking changes
• Monitoring for regressions in bias or truthfulness
• Documenting tradeoffs when perfect information is not available

Authority.inc expects to publish selected findings and frameworks from the partnership to help set a higher standard for AI-mediated procurement across the ecosystem, not just within its own products.

"In a year, we want to be able to say: this is not just a better ranking algorithm, this is a research-backed standard for how AI should behave when it touches vendor selection. That's the bar we're holding ourselves to as a research lab and as a product company."

About Oxbridge AIx

Oxbridge AIx is a premier AI entrepreneurship community focussed on Oxford Cambridge, connecting real-world AI deployments with research groups across the University of Oxford and the University of Cambridge. By bridging industry and academia, Oxbridge AIx supports projects focused on safety, ethics, and long-term societal impact. As part of this partnership, Oxbridge AIx is facilitating Authority.inc's collaboration with AI research groups, starting with our role as Research Sponsor of Oxford Artificial Intelligence Society and sponsoring research with their Safety and Governance Team.

About Authority.inc

Authority.inc is building the authority layer of the internet for B2B products and operating as an applied research lab on LLM search and ranking for procurement.

Starting with finance, fintech, insurance, and risk-related vendors, Authority.inc ingests, verifies, and publishes structured, machine-readable facts about companies and their offerings — including pricing, features, integrations, compliance, uptime, and more.

This data is exposed through LLM-optimized JSON-LD, public vendor catalogs, and APIs, enabling AI systems and human buyers alike to rely on verifiable, transparent context instead of opaque, ungrounded summaries.

Learn more at https://authority.inc