Discover TÜV-certified GoogleTest with Agentic AI for C/C++ testing!
Get the Details »
ISO/PAS 8800:2024
ISO/PAS 8800 introduces the first international specification for automotive AI safety. Parasoft helps OEMs and suppliers turn ISO/PAS 8800 into an engineering workflow by automating the verification, validation, and traceability that build your safety case for ADAS, SDV, and autonomous software.
ISO/PAS 8800:2024, Road vehicles — Safety and artificial intelligence, is the first international specification for the functional safety of AI in series-production road vehicles. It defines a framework for managing AI risk across the full development lifecycle. ISO/PAS 8800 doesn’t replace standards your team already uses. Instead, it builds on them.
Parasoft automates the verification, validation, and traceability that ISO/PAS 8800 requires for safety-critical software behind AI-enabled vehicle functions.
Engineering Safety for AI With ISO/PAS 8800
AI brings a new class of risk that traditional functional safety wasn’t built to catch. ISO 8800 gives the automotive industry a shared framework to manage it without throwing away ISO 26262 or SOTIF.
An AI element can work exactly as built and still be unsafe due to an insufficiency where it lacks the data or generalization to handle a new situation, like a partially hidden pedestrian.
ISO/PAS 8800 explicitly extends ISO 26262, ISO 21448 (SOTIF), and ISO/IEC 22989, adding the AI layer rather than replacing safety practices.
Training data coverage, balance, and edge case completeness each undergo a lifecycle of definition, verification, and validation.
AI can’t be proven perfectly safe, so you build a defensible safety case from evidence like robustness, coverage, architectural safeguards, and monitoring.
The work doesn’t stop at deployment: operational monitoring, field data, periodic reassessment, and controlled OTA updates keep the AI safe over the vehicle’s life.
ISO/PAS 8800 doesn’t stand alone. It sits on top of your functional-safety and SOTIF foundations and adds the AI-specific layer of assurance.
Extends ISO 26262 and SOTIF to AI/ML, addressing data quality, robustness, the assurance argument, AI-aware V&V, and operational monitoring.
Brings functional safety to E/E systems through the systematic and random-fault discipline your team practices at every ASIL.
Ensures safety of intended functionality by addressing performance limitations and foreseeable misuse of sensing and decision systems.
Governs vehicle lifecycle cybersecurity through threat modeling, risk assessment, and security controls that ISO/PAS 8800’s AI assurance depends on.
ISO/PAS 8800 stresses identifying AI risk early through a shift-left approach. Parasoft’s automotive suite verifies the safety-critical software surrounding every AI element and generates connected evidence to support ISO/PAS 8800 compliance.
Catch traditional software faults before they become AI-triggering conditions. 100% coverage of MISRA, AUTOSAR C++14, and CERT, plus CWE and OWASP security standards.
Where AI outputs feed safety-critical software, prove control logic, guardrails, and safety mechanisms behave correctly with TÜV SÜD–certified GoogleTest, backed by statement, branch, and MC/DC coverage.
Validate the connected services and zonal/SDV interfaces AI depends on, and simulate ECU and backend dependencies so testing never waits on hardware.
Bidirectional traceability links requirements to test cases, results, and code with centralized dashboards that produce the evidence for your assurance argument.
AI agents connect to static analysis, coverage, and standards data, autonomously fixing violations, closing coverage, and generating compliant justifications to keep pace with iterative AI safety work.
“Parasoft integrates seamlessly into our CI/CD pipeline. We’ve embedded test automation and static code analysis early in the development cycle — a huge win for quality assurance in safety-critical automotive projects.”
-Solero Technologies
See how Parasoft automates the verification, validation, and traceability that build your automotive AI safety case across both AI and traditional software.