At its 199th session in Geneva (June 23–26, 2026), WP.29 adopted a matched pair of instruments for Automated Driving Systems: a UN Regulation under the 1958 Agreement, and a Global Technical Regulation (GTR) under the 1998 Agreement, built with harmonized technical content across both. The US delegation voted for adoption, alongside China, the EU, Japan, and the UK — up to 62 parties.

Whether or not it's the law in your market yet, here's why it matters — and how SHM™ already delivers what it asks for.

That single vote hides two very different outcomes. For the EU, UK, Japan, and other 1958 Agreement parties, the UN Regulation becomes binding under international law once it enters into force. For the US, China, and India — parties to the 1998 Agreement only — the GTR obligates NHTSA to seriously consider domestic adoption through its own rulemaking process, but Article 7 of the 1998 Agreement explicitly permits a party to decline, as long as it publishes its reasons to WP.29. No violation, no penalty. Just a paper trail. So if you're a US-based AV company, you're probably not legally bound by this text. Not yet, and maybe not for a while, especially given NHTSA's recent deregulatory posture on FMVSS.

That doesn't make it irrelevant. If you sell into global markets, this is the bar your vehicle clears somewhere, bindingly, in the EU, UK, and Japan. And if you're operating in the US without a mandated bar, this is now the reference document insurers, plaintiffs' attorneys, and eventually regulators will reach for to ask "why didn't you" after the fact. A voluntary safety self-assessment is only as strong as what's inside it — and the industry just got a globally-endorsed description of what "strong" looks like.

The GTR asks manufacturers to build a safety case: evidence, not assertions, that a system is free of unreasonable risk, consistent across simulation, test track, and real roads, and proven again after deployment. All of it rests on one deceptively simple sentence: an ADS must perform at least as well as "a competent and careful human driver." Nobody defines that in measurable terms. Not the GTR, not anywhere else in the field.

That's one instance of a larger pattern: the regulation keeps asking for evidence — a human baseline, cross-pillar consistency, real-world monitoring — without supplying the instrument that would produce it. SHM™ was built for exactly that gap: a continuous, physics-based, assumption-free measure of collision hazard that means the same thing everywhere.

‍

Over the coming weeks, we’ll publish four pieces to walk through exactly where this applies:

- Defining "competent and careful," empirically | §3.1 - p. 22. The GTR's entire safety concept leans on that phrase without ever quantifying it. SHM™ already distinguishes skilled driving from unskilled driving, and establishes what "careful" looks like for a given ODD, directly from ordinary driving data — no need for millions of miles to find the signal. If the regulation's safety bar is a comparison to human driving, somebody has to be able to draw that baseline.

- Proving your evidence holds together across every test pillar | §6.3.1.4(d). Confirmatory testing and cross-pillar consistency only work with a measure that's the same everywhere it's computed (such as SHM™).

- Turning ordinary driving data into a scenario library the GTR already asks for | Annex 5. Annex 5 wants scenarios built from real data — we've been doing exactly this for transit corridors for years.

- Monitoring that proves a fix worked, not just that you're watching | §5.1.8.5.5(g). In-service monitoring is an obligation to show remedial action reduced risk, and that takes evidence, not a dashboard.

‍

Wherever your work sits — regulated today or not — this is worth twenty minutes of your time. Let's talk about what your safety case needs.

‍

\\\

‍

Original photo by John Samuel via Wikimedia

‍