Road safety data from fleets across the U.S., Canada, and Mexico suggests crash risk dropped in 2025, but danger on the road hasn’t gone away. That’s according to the 2026 Motive AI Road Safety Report, which looked at more than 1.2 billion hours of AI-detected safety events on commercial vehicles.

What’s Related

The report shows that for every one collision, fleets saw seven near-collisions, defined as situations where something almost went wrong but didn’t result in real damage. That pattern suggests fleets can now see danger early and act before a crash happens.

Early numbers also show fewer serious crashes in 2025. Severe collisions involving injuries, tow-aways, and deaths trended 9.5% lower, and reported injuries dipped 7.7% compared with the year before. That is close to the national traffic data, which shows an 8.2% drop in traffic deaths in the first half of 2025.

The report suggests the biggest shift isn’t traffic levels, but visibility into risk. AI data now helps fleets spot unsafe behavior earlier, before it turns into a crash.

“What’s different now is that AI allows organizations to see it earlier, coach drivers more consistently, and intervene before unsafe behavior turns into a collision,” said Hamish Woodrow, head of strategic analytics and data engineering at Motive. “This isn’t a temporary traffic lull or a seasonal dip — it reflects sustained changes in how organizations identify, coach, and manage risk.”

Why risk still matters

Even with fewer crashes overall, danger on the road still shows up in certain behaviors and times of day. The data shows:

• Late-night driving is risky. Crash risk is highest around 3 a.m., about three times higher than midday.
• Drowsy driving jumps early in the morning. Risk climbs quickly between 6 and 7 a.m.
• Distractions still matter. Cell phone use stays high in late afternoons, when drivers are tired, traffic is heavy, and delivery pressure peaks. Smoking behind the wheel also showed up thousands of times per day in the data.
• Aggressive driving plays a big role. Speeding, hard corners, and lane swerves were far more common among drivers involved in collisions.

The report also found that risk varies by location and industry. Driving behavior looks different in dense cities than in suburbs or rural areas, and sectors like waste and recycling or field services see more collisions than long-haul trucking.

What fleets can watch for in 2026

The report points to a shift in how safety is managed. Near-collision data, the signals that come before a real crash, is becoming more important than looking at collisions after they happen. The idea is that catching risk early gives fleets a chance to coach drivers and prevent damage.

“Motive’s dual-facing AI Dashcams have allowed us to approach safety in a much more proactive way,” said Paul Fly, director of risk and safety at Ernst Concrete. “In the past, we looked at lagging indicators and losses. Taking a proactive approach to safety has enabled us to reduce distracted driving and cell phone use by 81% in 13 months.”