NVIDIA DRIVE AV Raises the Bar for Car Security as Mercedes-Benz CLA Earns Prime Euro NCAP Award

AI-powered driver help applied sciences have gotten normal gear, basically altering how automobile security is assessed and validated.

The current recognition of the Mercedes-Benz CLA as Euro NCAP’s Finest Performer of 2025 underscores this shift, because the automobile combines conventional passive security options with NVIDIA DRIVE AV software program to realize the best total security rating of the yr.

??“When Euro NCAP assesses automobile security, it evaluates each passive and energetic techniques — attaining an ideal rating requires a state-of-the-art superior driver help system,” mentioned Ola Källenius, CEO of the Mercedes-Benz Group. “This milestone represents the fruits of 5 years of collaboration between Mercedes-Benz and NVIDIA to reinforce real-world security and ship tangible worth to clients.”

Euro NCAP (European New Automobile Evaluation Programme) has for practically 30 years served as Europe’s impartial automobile security authority, backed by European governments, motoring organizations and shopper teams.

Euro NCAP evaluates automobiles throughout 4 classes that mirror real-world security. For AI-powered driver help, probably the most related are the “Weak Highway Consumer” and “Security Help” classes, which assess applied sciences designed to assist forestall crashes — together with automated emergency braking, lane-keeping assist and velocity help.

Solely automobiles attaining five-star scores with normal gear qualify for “Finest in Class” recognition, with winners decided by weighted scores throughout all classes. In 2025, Euro NCAP examined a document 49 fashions.

Security Comes First: How DRIVE AV Is Constructed for Belief

Security scores like Euro NCAP are more and more recognizing automobiles that mix sturdy passive safety with superior energetic security efficiency. As AI turns into central to driving, the benchmark for the “most secure” automotive will more and more be outlined not solely by how properly a automobile handles a crash, however how successfully it helps forestall one.

The Mercedes-Benz CLA is constructed with NVIDIA DRIVE AV, a dual-stack structure that’s designed to assist automakers ship techniques that aren’t solely clever, however predictable, verifiable and resilient in the true world. The structure pairs an AI-driven end-to-end driving system with a parallel classical security stack to supply redundancy throughout AV sensing, planning and execution.

The CLA can also be constructed on the NVIDIA DRIVE Hyperion structure, which contains sensor variety and {hardware} redundancy into the automobile’s total design.

On the coronary heart of this method is NVIDIA Halos — a complete security system spanning {hardware}, software program, instruments, growth processes and certification assist. Halos delivers a structured security basis for growing automated driving and different AI capabilities whereas staying anchored to sturdy guardrails, redundancy and fault tolerance.

Third-party certification and assessments are additionally essential to construct belief:

• TÜV SÜD granted the ISO 21434 Cybersecurity Course of certification to NVIDIA for its automotive system-on-a-chip, platform and software program engineering processes. Moreover, NVIDIA DriveOS 6.0 conforms to ISO 26262 Automotive Security Integrity Stage (ASIL) D requirements.
• TÜV Rheinland carried out an impartial United Nations Financial Fee for Europe (UNECE) security evaluation of NVIDIA DRIVE AV associated to security necessities for advanced digital techniques, which NVIDIA efficiently accomplished.

NVIDIA not too long ago launched its Alpamayo household of open AI fashions, simulation instruments and datasets — which permits AVs to navigate even uncommon, “long-tail” occasions they haven’t been educated on by breaking the state of affairs down into smaller steps, reasoning via a number of potential actions and in the end choosing the most secure one. Utilizing these fashions with the parallel classical security stack within the NVIDIA DRIVE AV dual-stack structure supplies an extra layer of safety to maintain automobiles working inside secure boundaries.

Coaching Security Via Information and Simulation

Fashionable AI-driven security techniques be taught from exponentially extra driving eventualities than any human may expertise in a lifetime. NVIDIA’s cloud-to-car growth method transforms real-world knowledge into billions of simulated miles utilizing NVIDIA DGX techniques for neural community coaching, the NVIDIA Omniverse and Cosmos platforms for simulation, and NVIDIA DRIVE AGX for in-vehicle computing.

This technique addresses a crucial problem in security validation: coaching AI to navigate uncommon however high-risk edge instances which are too harmful — or too rare — to check reliably in the true world. By producing artificial eventualities that signify these uncommon conditions, AI techniques can be taught applicable responses throughout growth with out placing folks in danger.

The CLA’s recognition is greater than a single mannequin incomes a prime ranking — it displays a broader shift in what security means within the trendy automobile, the place trusted crash safety is paired with AI-enabled driver help designed to assist keep away from accidents within the first place.