AES-GCM

CryptoLab: Interactive Educational Platform Turns Cryptographic Protocols Course into Hands-On Alice, Bob and Mallory Experiments

A university instructor developed CryptoLab, an educational testbed that lets students run live cryptographic protocol experiments with active attackers instead of simple encrypt-decrypt exercises. The platform models the classic Alice-Mallory-Bob scenario where each participant runs as a separate application mode, allowing inspection, modification, dropping, and replay of packets. The first lab focuses on AES-GCM, AEAD properties, nonce reuse, replay protection, weak RNGs, key rotation, and metadata leakage. Students can operate in interactive mode to manually attack traffic or execute automated scenarios that verify expected security outcomes. Experiments demonstrate that modifying ciphertext triggers authentication failure and that a valid authentication tag does not guarantee message freshness without additional replay defenses. The tool also includes Python 3.10 assignments for correctly using cryptographic primitives rather than implementing algorithms from scratch. CryptoLab currently supports Windows x64 and macOS Apple Silicon builds and serves as the foundation for upcoming labs on symmetric encryption modes, asymmetric cryptography, key exchange, and TLS.

Claude Encrypted Thinking Blocks Use Protobuf with Exposed Metadata and AES-GCM Ciphertext

A detailed reverse-engineering of Claude signatures shows that the encrypted reasoning blocks are not opaque containers but structured protobuf messages. The outer envelope contains a 312-byte inner message that holds a 135-byte header, fixed-length nonce and MAC fields, and the actual ciphertext. The header itself reveals the model name such as claude-opus-5, the block type as thinking, and the organizationUuid from the user's Anthropic account. Only the reasoning text is encrypted with AES-GCM, adding exactly 16 bytes for the authentication tag. The analysis covers four protocol versions and notes that organization binding was added in version 15, potentially allowing servers to reject cross-model or cross-organization reuse. The findings provide concrete implications for both the Opus-to-Haiku extraction attack and the leakage of account identifiers in public logs.