Darkmatter Exo 4.7 Beta 2 May 2026

The biggest change under the hood is the swap from CRYSTALS-Kyber to FrodoKEM-976. Why? Kyber relies on Module-LWE (Learning with Errors), which, while efficient, has a smaller security margin against side-channel attacks when implemented in software. FrodoKEM is based on standard LWE, which is mathematically “cleaner” and avoids the complicated structure that some researchers fear could be exploited by a future quantum algorithm.

Performance trade-off: In Beta 2, key exchange is approximately 40% slower than in 4.6. Handshake latency for a new session jumps from 120ms to ~210ms. Darkmatter’s dev notes argue this is acceptable for high-security modes and will be optimized by the Release Candidate (RC). Darkmatter Exo 4.7 Beta 2

This is the "wow" feature—or the "why" feature, depending on your threat model. ANI monitors timing patterns, power fluctuations, and even CPU cache hits. When it detects a potential side-channel probing attempt (e.g., a Flush+Reload attack), it injects decoy cryptographic operations and synthetic noise into the pipeline. The biggest change under the hood is the

Beta 2 quirk: ANI is overzealous. Testers have reported false positives when running Exo inside noisy virtualized environments (VMware, QEMU) or high-load Kubernetes pods. Darkmatter has included a --ani-threshold flag (1–10, default 5) to tune sensitivity. FrodoKEM is based on standard LWE, which is

The external chassis utilizes the second generation of Reactive Darkmatter Composite. This material does not merely absorb radar waves; it actively disperses sensor pings into the exo’s heat sinks.