Whitepapers

Flagship papers1 PDFs availableABIS archive
The formal publicationlayer for the ABIS research run.

Eleven research papers covering the full ABIS program — from deterministic drift detection to behavioral geography — with summaries, metadata, and linked PDFs where available.

Read featured papersSee dated findingsOpen ABIS program

Archive status

The public paper set now reads like a coherent release, not a loose list.

Eleven papers spanning deterministic drift detection, cross-domain transfer, instantaneous measurement, comparative stability, energy distance, correction limits, mathematical frameworks, behavioral rigidity, catastrophic fragility, behavioral geography, and feature hierarchy — all in one visible archive with metadata, summaries, and PDF assets where available.

11 total papers1 published1 downloadable PDFs

Summary

every page

Metadata

always visible

Public

citation layer

Latest paper

Deterministic Detection of Large Language Model Behavioral Drift

A deterministic approach to detecting LLM behavioral drift using information-theoretic and mathematical text signatures, without training data, learned monitors, or baseline distributions.

Open 10 Mar 2026

Bridge to findings

The papers connect to dated public releases.

Whitepapers state the claims. Findings studies show how those ideas land in monitored model behavior over time.

Open findings studies

11

papers in archive

11

featured now

1

public PDFs

1

published papers

Archive structure

Each paper page includes:

  • A plain-English summary of the paper's claim.
  • Metadata for date, proof level, and manuscript status.
  • A PDF link when a public asset is attached.

Current emphasis

The archive now contains eleven papers covering deterministic drift detection, cross-domain transfer, instantaneous measurement, comparative model stability, energy distance analysis, closed-loop correction, mathematical frameworks, behavioral rigidity, catastrophic fragility, behavioral geography, and dual-tier feature hierarchy.

Open the ABIS program page

See dated findings studies

Program context

ABIS

Read the program framing first if you want the publication archive to sit inside the broader measurement story.

Open program

Library view

Research library

Step back from the papers to see how notes, findings studies, and operational scenarios connect to the archive.

View library

Observed behavior

Drift monitor studies

Move from formal claims into dated releases built around monitored behavior and benchmark snapshots.

Read studies

Full Archive

All 11 papers in the public library.

Each paper has a summary page, metadata, and where available a linked PDF. The site supports manuscripts even when a public PDF is not yet attached.

Published paper

Deterministic Detection of Large Language Model Behavioral Drift

A deterministic approach to detecting LLM behavioral drift using information-theoretic and mathematical text signatures, without training data, learned monitors, or baseline distributions.

10 Mar 2026validated
Open summary

Research manuscript

Dual-Tier Feature Hierarchy in Behavioral Drift Detection

Feature-importance analysis across two training regimes suggests a hierarchy of universal indicators and domain-specific specialists in behavioral drift detection.

5 Mar 2026validated
Open summary

Research manuscript

A Behavioral Geography of Large Language Models

A 29-model mapping study that treats LLM behavior as a measurable geography in high-dimensional feature space rather than as a simple ranking problem.

22 Feb 2026validated
Open summary

Research manuscript

Catastrophic Fragility in Sequential Drift Detection

A case study showing catastrophic training collapse in a Transformer-based drift detector, contrasted with successful XGBoost performance on the same task.

18 Jan 2026validated
Open summary

Research manuscript

Behavioral Rigidity Spectrum

A quantitative study of how resistant different frontier LLMs are to prompt-level correction once drift has been detected.

20 Oct 2025validated
Open summary

Research manuscript

Closed-Loop Deterministic Correction of AI Behavioral Drift

A closed-loop system that detects behavioral drift and attempts escalating correction, revealing a boundary between correctable and architecturally irreversible change.

15 Aug 2025validated
Open summary

Research manuscript

Comparative Behavioral Stability of Large Language Models

A 37-model study that compares frontier model stability using deterministic behavioral analysis rather than capability-only evaluation.

18 Jun 2025validated
Open summary

Research manuscript

Energy Distance as a Universal Indicator of Behavioral Collapse

A paper arguing that Energy Distance behaves like a strong universal drift indicator, especially in cross-domain transfer settings.

11 Apr 2025validated
Open summary

Research manuscript

Behavioral Drift Is Instantaneous, Not Sequential

Evidence that sufficiently rich behavioral representations make drift detectable from a single observation window rather than requiring long sequential context.

14 Feb 2025validated
Open summary

Research manuscript

Universal Behavioral Drift Has a Mathematical Signature

A cross-domain paper arguing that behavioral drift shares a measurable mathematical signature across fundamentally different systems.

12 Dec 2024validated
Open summary

Research manuscript

Information-Geometric and Topological Features of AI Behavioral Signatures

A paper describing the 14-module mathematical framework used to construct high-dimensional behavioral signatures from raw outputs.

15 Nov 2024validated
Open summary