TIER 1 · ACTIVE PURSUIT · AVIATION AND AIR TRAFFIC AI

Audit infrastructure /
for aviation AI.

FAA and EASA are actively defining AI audit requirements for aviation. DO-178C governs avionics software today, including AI/ML components in certified systems. Aviation has the highest per-event liability of any segment after nuclear — every AI-mediated flight incident becomes litigation. EASA jurisdiction extends to any operator flying EU routes: Delta ($1.74B EU AI Act exposure), Lufthansa (€1.14B), AF-KLM (€0.93B), IAG (€0.84B) all face EASA compliance requirements regardless of FAA framework. The audit infrastructure must be in place before the incident, not after.

FAA AI/ML Aviation Safety Policy (2024): AI used in Type Certificated products requires documented performance monitoring and audit logging.
FAA AI/ML Policy
Active · US
AI/ML Aviation Safety Policy (2024)
FAA 2024 policy on AI/ML in aviation explicitly requires documented performance monitoring, drift detection, and audit logging for AI systems used in Type Certificated products and air traffic management. The policy is operational — not proposed.
EASA AI Roadmap
Active · EU
AI Roadmap 2.0 (2024)
EASA AI Roadmap 2024 defines a framework for AI use in aviation including Level 1 and Level 2 AI applications with proportionate assurance requirements. Higher assurance levels require documented, auditable AI decision logs.
DO-178C
Active · RTCA
Software considerations in airborne systems
DO-178C is the current and only operative airborne software standard — RTCA has not announced a successor. Its verification, traceability, and configuration management requirements apply to AI/ML components in Type Certificated systems today. Organizations embedding AI in certified products must satisfy DO-178C traceability requirements now, not at some future date.
ICAO AI Task Force
Active development
Global AI aviation framework
ICAO AI Task Force is developing global recommendations for AI in aviation. The framework converges on tamper-evident decision logs as a safety and accountability requirement. ICAO recommendations typically become national regulation within 3–5 years.
SAE ARP6983 / EUROCAE ED-324 — AI/ML Aviation Standard
Publication expected 2026–2027
SAE ARP6983 / EUROCAE ED-324 — process standard (in development)
SAE ARP6983 (jointly EUROCAE ED-324) is the dedicated forthcoming process standard for developing and certifying AI/ML in avionics, written by SAE committee G-34 with EUROCAE WG-114 — not RTCA, which publishes the DO-xxx series. It is still in development (publication targeted ~2026); Issue 1 is scoped to frozen, supervised ML up to Design Assurance Level C (DAL C), with extensive verification, validation, and traceability documentation. Organizations pre-building audit infrastructure before publication avoid the cost and scramble of retroactive certification. DO-178C remains the operative standard until publication.
EASA CS-25 — Large Aeroplanes
Active · EU + global reach
Certification Specification — EU type certificate authority
EASA CS-25 is the EU equivalent of FAA Part 25 for large transport aircraft type certification. AI/ML components in CS-25 aircraft require the same documentation and traceability as DO-178C systems. Critical jurisdictional reach: EASA authority applies to any operator flying EU routes regardless of aircraft registration — giving EASA AI compliance requirements effective global application for any carrier with EU operations.
DO-326A / DO-356A
Active · RTCA
Airworthiness Security Process + Methods — cybersecurity
DO-326A (airworthiness security process) and DO-356A (airworthiness security methods and considerations) are mandatory for avionics systems with network connectivity. AI systems receiving inputs from connected sources must satisfy these security standards. Tamper-evident audit records of AI decision inputs and outputs satisfy the security evidence requirements for AI-enabled airborne systems.
NTSB / AAIB / BEA / BFU / JTSB
Active · international
Accident investigation — 5-jurisdiction authority
Aviation accident investigators in the US (NTSB), UK (AAIB), France (BEA), Germany (BFU), and Japan (JTSB) have subpoena authority over all AI decision records relevant to an accident. Tamper-evident audit chains from the moment of deployment — not reconstructed post-accident — are the gold standard for exculpatory or liability evidence. Boeing 737 MAX 8 ($20B+ cumulative settlements partly due to disputed MCAS decision records) established the precedent for how AI decision records translate to settlement outcomes.

* EU AI Act Annex III enforcement date: August 2, 2026 (legally operative). EU Digital Omnibus provisional agreement (May 7, 2026) proposes extending to December 2, 2027 — not yet formally enacted. Prepare for the earlier date.

PRIMITIVE
REQUIREMENT SATISFIED
AUDIT CHAIN
FAA AI/ML Policy · DO-178C compliance evidence
Every AI aviation decision — flight path optimization, departure sequencing, maintenance anomaly classification — logged in a tamper-evident, hash-chained record. Independently verifiable by FAA, EASA, or NTSB investigators without relying on the operator or manufacturer.
IDENTITY REGISTRY
Avionics system version traceability
Each aviation AI system version has a verifiable, traceable identity on Base mainnet. Software updates are documented events. In accident investigation, the exact system version responsible for any decision is unambiguously established — satisfying DO-178 configuration management requirements.
DECISIONAL LOGGING
Per-decision capture at high-stakes tier
High-stakes tier for safety-critical decisions: conflict detection, emergency descent, systems failure response. Consequential tier for flight planning and sequencing decisions. IPFS-pinned evidence payloads include sensor state and decision inputs.
AUDIT CHAIN AND IDENTITY REGISTRY ARE LIVE ON BASE MAINNET TODAY.
ROUTINE
$0.0001 / event
Routine events
CONSEQUENTIAL
$0.01 / event
Consequential events
HIGH-STAKES
$0.10 / event
High-stakes events
VOLUME NOTE
LCC / regional carrier (1,000–5,000 flights/month): 100 high-stakes × 3,000 flights × $0.10 = $30K/month + consequential layer ~$50K/month → ~$1M/year. Major carrier (Delta, Lufthansa, AF-KLM tier, 60K flights/month): high-stakes alone $600K/month, plus consequential $1M–$5M/month, plus routine $200K–$500K/month → $5M–$15M/year. Per-aircraft annual alternative: $5K–$25K/aircraft/year flat for routine + consequential bundled; high-stakes metered.
SCALE
LCC / regional carrier: $250K–$1M/year. Major carrier (Delta $58B revenue, Lufthansa €38B): $3M–$20M/year. ANSP (FAA ATO, NATS, DFS, Eurocontrol, NavCanada): $2M–$15M/year. Avionics vendor (Honeywell, Collins Aerospace, Thales, Safran): $500K–$5M/year. Aircraft OEM (Boeing, Airbus, Embraer): $5M–$25M/year + per-aircraft royalty. MRO operator (Lufthansa Technik, AAR, ST Engineering): $250K–$3M/year per facility. eVTOL / UAS (Joby, Archer, Wisk, Zipline): $100K–$2M/year.
WHO BUYS THIS
Boeing · Airbus · Honeywell Aerospace · Collins Aerospace (RTX) · Thales Avionics · NAVBLUE (Airbus) · NATS (UK ANSP) · Eurocontrol · Delta Air Lines · Lufthansa · American Airlines · United Airlines · Southwest Airlines · Air France-KLM · IAG (British Airways / Iberia) · Emirates · Singapore Airlines · Cathay Pacific · LATAM Airlines · Turkish Airlines · Wizz Air · Ryanair · easyJet · Joby Aviation · Archer Aviation · Wisk Aero · Zipline · Vantis (BNSF / Rockwell) · FAA Air Traffic Organization · Japan JCAB · Transport Canada Civil Aviation · Australia CASA · Brazil ANAC · India DGCA · Saudi GACA · UAE GCAA
WHAT THIS REPLACES
Boeing 737 MAX 8 (2018–2019): $20B+ cumulative settlements partly due to disputed MCAS software decision records. EASA 3% global turnover penalty: Delta $1.74B, Lufthansa €1.14B, AF-KLM €0.93B, IAG €0.84B exposure. Single major AI-mediated incident post-2027: $5B–$30B+ litigation exposure based on MAX 8 precedent. DO-178C re-verification on AI/ML component: $1M–$10M per certification cycle. Avionics certification delay due to inadequate AI documentation: $50M–$500M in delayed launch revenue.
ACTIVATION TRIGGER
August 2, 2026: EU AI Act Annex III high-risk obligations become operative for aviation AI. EASA's dedicated AI assurance framework is forthcoming — proposed via NPA 2025-07, with finalisation expected later this decade — rather than a fixed audit start date. FAA AI/ML Aviation Safety Policy (2024) is operative now — AI in Type Certificated products requires documented performance monitoring and audit logging. SAE ARP6983 / EUROCAE ED-324 (the dedicated AI/ML process standard, up to DAL C) is in development, publication expected ~2026. EASA jurisdiction extends to any operator flying EU routes — every major US carrier faces EASA AI compliance regardless of FAA framework.
MULTINATIONAL DEPLOYMENT

EASA jurisdiction extends structurally to any carrier flying EU routes regardless of aircraft registration — making EASA AI compliance effectively global for any operator with EU operations. A single Yolo deployment produces the record-keeping evidence FAA AI/ML Policy (US), EASA CS-25/AI Roadmap (EU), Transport Canada TCCA AI guidance, JCAB (Japan), and CASA (Australia) each require.

ENTERPRISE INQUIRIES

For director of safety / chief safety officer, avionics DER, head of AI/ML certification, type certificate holder regulatory liaison, FOQA program managers, cybersecurity airworthiness lead (DO-326A/356A), MRO Part 145 quality manager.

agents@yolo.solutions

See the full compliance overview at yolo.solutions/compliance and the developer integration guide at /developers/decisional-logging.

← ALL SEGMENTS