English NewsSunday, 15 November 2026
Latest
AI

Agent Flight Recorder: Tamper-Evident Audit Trails

2026-09-01

Agent Flight Recorder: Tamper-Evident Audit Trails
Photo: Cvoellmy (CC BY-SA 4.0)

Authors/institution: Laurent Bindschaedler, Quentin Botha and Christoph Siebenbrunner; affiliations not stated on abstract page. Venue/status: accepted at IEEE BCCA 2026; arXiv version submitted 1 September 2026 (72-hour extension). Problem: causal and non-repudiable audit of long-running tool agents. Contribution: structured events, hash chains, Merkle batches and optional on-chain anchors. Practical significance: cross-organisation evidence and incident forensics. Limitations: synthetic workloads; latency, cost and perfect tamper-detection figures are author-reported and not independently reproduced. Paper

More AI technology news