AI Used to Decrypt Medieval Ciphers

2026-06-03 · Source: Schneier on Security · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Research Methodology & Innovation · Depth: Novice, short

Summary

Researchers are employing machine learning algorithms to decrypt historical medieval ciphers, specifically targeting an estimated 1% of archival material that remains fully or partially encrypted. The DESCRYPT project utilizes both the online platform Transkribus, which hosts over 300 community AI models, and custom-developed AI tools to reveal messages from these ancient documents. This approach capitalizes on AI's statistical capabilities to invert plaintext statistics from ciphertext, effectively multiplying them back to the original message. While promising, the process faces challenges where context is crucial, as highlighted by examples like O.Henry's "Calloway's Code" and Cockney Rhyming Slang, which rely on nuanced understanding beyond mere statistical likelihoods. The initiative aims to reveal centuries-old secrets from these complex historical encryptions.

Key takeaway

For research scientists focused on historical cryptography or archival preservation, this AI application demonstrates a powerful new approach. You should explore integrating statistical AI models, potentially via platforms like Transkribus, to tackle the estimated 1% of unread encrypted archival material. Be mindful that while AI excels at statistical inversion, complex ciphers requiring deep contextual understanding may still pose significant challenges, necessitating hybrid human-AI methods.

Key insights

AI's statistical power can invert ciphertext statistics to decrypt historical ciphers, revealing hidden archival content.

Principles

• Ciphers flatten plaintext statistics.
• AI inverts ciphertext statistics.
• Context is vital for decryption.

Method

Researchers use community AI models via Transkribus and custom AI tools to statistically invert ciphertext, revealing partially or fully encrypted historical documents.

In practice

• Decrypt historical documents with AI.
• Use Transkribus for archival material.
• Develop custom AI decryption tools.

Topics

• AI Decryption
• Medieval Ciphers
• Historical Cryptography
• Machine Learning Applications
• Transkribus
• Archival Preservation

Best for: NLP Engineer, AI Scientist, Research Scientist, AI Student

Related on AIssential

• Turbo Pascal 3.02A, deconstructed — AI models can effectively deconstruct and interpret legacy binary executables into human-readable code.
• An AI agent hacked McKinsey's internal AI platform in two hours using a decades-old technique — Classic SQL injection vulnerabilities can enable silent, widespread AI system manipulation via prompt and RAG data access.
• Roots Beneath the Cut: Uncovering the Risk of Concept Revival in Pruning-Based Unlearning for Diffusion Models — Pruning-based unlearning in diffusion models is vulnerable to concept revival via side-channel information from pruned weight locations.
• Archivists Turn to LLMs to Decipher Handwriting at Scale — General-purpose LLMs now reliably transcribe historical handwriting, outperforming specialized software in speed, cost, and accuracy.
• With $1 Cyberattacks on the Rise, Durable Defenses Pay Off — Generative AI accelerates both cyberattacks and defenses, necessitating a shift to foundational security rather than reactive patching.
• Shafi Goldwasser Provides 'A Cryptographic Perspective on Trustworthy AI' — Cryptography offers a robust framework for defining, achieving, and proving trustworthiness in AI systems against adversarial threats.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Schneier on Security.