New York – May XX, 2026 – Lockton, the world’s largest privately held insurance brokerage, and Nexar, the real-world intelligence platform for the Physical AI era, today introduced a human-benchmark framework for evaluating autonomous vehicle (AV) safety against real-world human driving behavior. The framework addresses a longstanding gap in the AV ecosystem: the absence of an independent, widely recognized standard for measuring AV safety relative to how humans actually drive.

Why a Human Benchmark Matters

Without a standardized benchmark, insurers, regulators, and AV companies face structural challenges. Insurers struggle to differentiate risk across AV systems, leading to blunt underwriting approaches or restrictions on capacity. Regulators lack a consistent reference for assessing safety claims. AV developers have no third-party-validated way to demonstrate performance against the standard the public most readily understands.

“For autonomous vehicles to scale, safety has to be measured against a credible, independent benchmark built on real-world driving. This framework provides that benchmark,” said Preet Gill, EVP and Leader of Lockton’s Global Technology Risk Practice. “As autonomy is adopted across logistics, mobility, and commercial fleets, even insurers who don't directly underwrite AVs will carry exposure, and our role is to help evaluate that risk before it becomes systemic.”

An Independent Framework, Built on Real-World Ground Truth

The framework brings together Lockton’s risk management and insurance expertise and Nexar’s real-world driving intelligence to answer a fundamental question: does an AV system perform at, above, or below human-level safety in comparable conditions?

It is designed as an open, evolving reference rather than a proprietary rating and is built to serve the three constituencies that need a credible answer to that question: insurers structuring coverage, regulators evaluating safety claims, and AV developers seeking third-party-validated ways to demonstrate performance.

“For the first time, AV systems can be benchmarked against a standard that wasn’t built by anyone with a stake in the outcome,” said Jon Miller, Chief Business Officer at Nexar. “That is what credible verification requires, and it is what this framework makes possible.”

The Data Foundation

The framework is powered by BADAS 2.0, Nexar's collision anticipation model family. Trained entirely on real-world driving, 10 billion miles and 60 million safety-critical events, with no synthetic data, BADAS 2.0 achieves 99.4% average precision and ranks #1 on all four major industry benchmarks. Nexar's network of 350,000 cameras across 94% of U.S. roads provides the scale required to establish the human baseline.

Initial Framework Components

The framework includes two initial Nexar capabilities supporting risk-based evaluation:

• Nexar Risk Index — an environmental risk assessment that evaluates the operating domain of an AV based on geography, conditions, and driving context, establishing contextual risk before system comparison.
• Nexar Apex — a submission platform that allows AV companies to verify their systems against curated real-world edge cases specific to their operating environments, with results indicating performance relative to the human driving baseline.

Built on Established Practice

Nexar data has already been used by leading AV developers to support published safety cases and peer-reviewed research. This framework formalizes approaches already in use at the highest levels of industry scrutiny, with Lockton’s risk governance and Nexar’s data infrastructure together extending those practices to the broader market.