Andrii Bidochko

• Updated: March 19, 2026
• 5 min read

NHTSA Intensifies Probe into Tesla’s Full Self‑Driving Supervised Software

Tesla FSD Supervised Investigation: What the NHTSA Probe Means for Autonomous Vehicles

The U.S. National Highway Traffic Safety Administration (NHTSA) has upgraded its probe into Tesla’s Full Self‑Driving (Supervised) software to an engineering analysis, a step that often precedes a mandatory recall, after uncovering multiple low‑visibility crashes where the system failed to alert drivers.

NHTSA intensifies scrutiny of Tesla’s FSD Supervised

In March 2026, the NHTSA’s Office of Defects Investigation (ODI) announced that it is deepening its investigation into Tesla’s Full Self‑Driving (Supervised) (FSD) software, specifically targeting incidents that occurred under reduced visibility conditions such as heavy rain, fog, and night‑time glare. The agency has now classified the probe as an “engineering analysis,” the highest level of technical review, signaling that a recall could be on the horizon if systemic flaws are confirmed.

Low‑visibility crash data: what the numbers reveal

Since the initial probe launched in October 2024, ODI has catalogued more than 80 reported violations of basic traffic laws—including red‑light runs—and four fatal or serious injuries linked to low‑visibility scenarios. The most alarming case involved a pedestrian struck at night when the FSD system failed to detect a degraded camera state.

• Four crashes in fog or heavy rain where the system did not issue a driver alert.
• Two incidents where the software lost track of a lead vehicle within 2 seconds of visibility loss.
• Multiple red‑light violations recorded in low‑light urban environments.

ODI’s analysis indicates that the FSD software often “did not detect common roadway conditions that impaired camera visibility” and, when it finally recognized the issue, the alert arrived after the collision point, leaving insufficient reaction time for the driver.

Tesla’s response and the road to a possible recall

Tesla acknowledges the investigation and points to an internal update begun in June 2024 aimed at improving camera‑degradation detection. However, the company has not yet confirmed whether the patch has been rolled out to the affected fleet, nor has it disclosed which VINs received the fix.

In a recent statement, Tesla’s Autopilot team emphasized that “continuous data collection and over‑the‑air updates are core to our safety strategy,” but ODI reports that the regulator has not received all requested data, raising concerns about under‑reporting of similar incidents.

What a recall could mean for Tesla and the industry

Should the NHTSA conclude that the FSD Supervised software poses an unreasonable risk, a recall could affect up to 1.2 million vehicles equipped with the latest hardware suite. The financial and brand impact would be significant, especially as Tesla pushes its robotaxi service in Austin, Texas.

A recall would also trigger a cascade of compliance requirements:

1. Mandatory software patches delivered via OTA updates.
2. Enhanced driver‑monitoring protocols for low‑visibility operation.
3. Re‑evaluation of the “Supervised” designation under federal guidelines.

Broader impact on autonomous vehicle safety standards

The investigation arrives at a pivotal moment for the autonomous vehicle (AV) ecosystem. Regulators worldwide are watching the NHTSA’s methodology as a benchmark for future AV safety assessments. Key implications include:

Companies developing AI‑driven driver assistance, such as OpenAI ChatGPT integration and ChatGPT and Telegram integration, are closely monitoring the outcome. A recall could accelerate the shift toward more robust, multimodal AI models that fuse vision, lidar, and radar data.

Read the original report

For a detailed account of the investigation, see the TechCrunch article:
Tesla FSD Supervised under NHTSA scrutiny.

How AI platforms like UBOS are shaping the future of vehicle safety

While regulators tighten standards, AI‑centric platforms are providing developers with the tools needed to meet them. The UBOS platform overview showcases a modular architecture that supports real‑time sensor fusion, essential for low‑visibility detection.

Startups can accelerate prototyping with UBOS templates for quick start, including a pre‑built AI SEO Analyzer that helps automotive SaaS products stay visible in a crowded market.

For SMBs looking to embed advanced analytics, the UBOS solutions for SMBs offer a low‑code Workflow automation studio that can trigger alerts when sensor data falls below confidence thresholds.

Enterprises benefit from the Enterprise AI platform by UBOS, which integrates with voice technologies like the ElevenLabs AI voice integration to provide audible warnings to drivers in real time.

Marketing teams can leverage AI marketing agents to promote safety‑focused features, while the UBOS partner program enables collaborations with OEMs and Tier‑1 suppliers.

Developers interested in data storage solutions can explore the Chroma DB integration, which offers vector‑based retrieval for rapid image‑and‑sensor query processing—crucial for low‑visibility scenario simulations.

For those building conversational assistants for in‑vehicle use, the AI Chatbot template and the GPT-Powered Telegram Bot provide ready‑made pipelines that can be fine‑tuned on driving‑specific corpora.

Cost considerations for integrating advanced safety AI

Companies evaluating a shift toward higher‑fidelity safety stacks should review the UBOS pricing plans. Tiered pricing allows startups to start with a free tier for prototyping, while enterprises can negotiate custom SLAs for large‑scale OTA deployments.

Bottom line

The NHTSA’s upgraded investigation into Tesla’s Full Self‑Driving (Supervised) software underscores a growing regulatory appetite for concrete safety evidence, especially under adverse conditions. While Tesla works to patch its vision‑based shortcomings, the broader autonomous‑vehicle ecosystem is likely to accelerate the adoption of multimodal AI, robust OTA frameworks, and transparent reporting mechanisms.

For developers, engineers, and product leaders, the lesson is clear: safety‑first AI architectures are no longer optional—they are a regulatory prerequisite. Leveraging platforms like UBOS can provide the modular, compliant foundation needed to stay ahead of both the market and the regulators.

Andrii Bidochko

CTO UBOS

Andrii Bidochko is an AI entrepreneur and researcher focused on AI agents, reinforcement learning, and autonomous systems. He writes about the technologies shaping the future of machine intelligence, from frontier models and agent architectures to real-world AI applications.