Walter Isaacson’s Elon Musk biography is set to be published on Tuesday, and a new preview of the book illustrates details about Tesla’s development of the upcoming Full Self-Driving (FSD) version 12.
In an additional preview of his Musk biography for CNBC, Isaacson discusses the use of AI in the development of Tesla’s FSD v12, in a shift that took place within the last several months. Isaacson talks about Tesla’s recent development of the upcoming FSD v12, which he and Tesla demonstrate has moved away from a “rules-based” approach.
Notably, FSD v12 is expected to use billions of video frames from real-world driving incidents to train its neural network rather than using thousands of lines of code like previous versions. In a conversation with Musk last December, Tesla Autopilot employee Dhaval Shroff had likened the concept to the popular chatbot ChatGPT, instead for use with driving.
“It’s like ChatGPT, but for cars,” Shroff said. “We process an enormous amount of data on how real human drivers acted in a complex driving situation, and then we train a computer’s neural network to mimic that.”
Surprisingly enough, Tesla only shifted toward this “neural network planner” approach recently. By the beginning of this year, however, the neural network had already analyzed 10 million video clips based on the best-case-scenario drivers the system had access to. Musk instructed employees at the company’s Buffalo, New York facility who were in charge of analyzing the footage to train the AI on things “a five-star Uber driver would do.”
Moving from a rules-based to a network-path-based AI approach allowed FSD to use human driving data to avoid obstacles, even if breaking some rules was necessary. Shroff helped demonstrate the idea to Musk with a demo featuring trash bins, debris, and upturned traffic cones, which the car handled surprisingly well.
“Here’s what happens when we move from rules-based to network-path-based,” Shroff explained. “The car will never get into a collision if you turn this thing on, even in unstructured environments.”
Musk quickly took to the idea, as can be seen in a recent livestream of Tesla’s FSD v12 software in Palo Alto with Autopilot software director Ashok Elluswamy. He has repeatedly spoken about the upcoming software version’s impressive driving results, despite one small moment in the drive where the car almost ran a red light.
In any case, Musk could argue that the red-light moment is a good case for the need for self-driving software to continually learn. Given that it will constantly be trained from the video data generated by camera footage from real-world drivers, it should theoretically make it safer over time, according to Musk.
During development, Musk also reportedly latched onto the fact that it took over a million video clips for the neural network to begin performing well, though he looks forward to what significantly more data will do for FSD.
Still, critics and regulators have expressed concerns about the faults of human drivers training AI-based driving systems, and Tesla has repeatedly been questioned by the National Highway Traffic Safety Administration (NHTSA) about its Autopilot and FSD beta systems.
According to Isaacson, Tesla plans to release FSD v12 as soon as regulators approve it. Meanwhile, an ongoing study by the National Highway Safety Board is looking to determine if self-driving cars should be permitted to imitate human driving actions that blur traffic rules, such as creeping up at stop signs.
Musk said in April that he expects Tesla to reach full autonomy within a year, though he has also been known to share ambitious targets for the software in the past.
You can read Walter Isaacson’s full account of the development of Tesla FSD v12 here, in a CNBC preview of the upcoming Elon Musk biography.
What are your thoughts? Let me know at zach@teslarati.com, find me on X at @zacharyvisconti, or send your tips to us at tips@teslarati.com.
News
Elon Musk’s Grok AI to be used in U.S. War Department’s bespoke AI platform
The partnership aims to provide advanced capabilities to 3 million military and civilian personnel.
The U.S. Department of War announced Monday an agreement with Elon Musk’s xAI to embed the company’s frontier artificial intelligence systems, powered by the Grok family of models, into the department’s bespoke AI platform GenAI.mil.
The partnership aims to provide advanced capabilities to 3 million military and civilian personnel, with initial deployment targeted for early 2026 at Impact Level 5 (IL5) for secure handling of Controlled Unclassified Information.
xAI Integration
As noted by the War Department’s press release, GenAI.mil, its bespoke AI platform, will gain xAI for the Government’s suite of tools, which enable real-time global insights from the X platform for “decisive information advantage.” The rollout builds on xAI’s July launch of products for U.S. government customers, including federal, state, local, and national security use cases.
“Targeted for initial deployment in early 2026, this integration will allow all military and civilian personnel to use xAI’s capabilities at Impact Level 5 (IL5), enabling the secure handling of Controlled Unclassified Information (CUI) in daily workflows. Users will also gain access to real‑time global insights from the X platform, providing War Department personnel with a decisive information advantage,” the Department of War wrote in a press release.
Strategic advantages
The deal marks another step in the Department of War’s efforts to use cutting-edge AI in its operations. xAI, for its part, highlighted that its tools can support administrative tasks at the federal, state and local levels, as well as “critical mission use cases” at the front line of military operations.
“The War Department will continue scaling an AI ecosystem built for speed, security, and decision superiority. Newly IL5-certified capabilities will empower every aspect of the Department’s workforce, turning AI into a daily operational asset. This announcement marks another milestone in America’s AI revolution, and the War Department is driving that momentum forward,” the War Department noted.
News
Tesla FSD (Supervised) v14.2.2 starts rolling out
The update focuses on smoother real-world performance, better obstacle awareness, and precise end-of-trip routing, among other improvements.
Tesla has started rolling out Full Self-Driving (Supervised) v14.2.2, bringing further refinements to its most advanced driver-assist system. The new FSD update focuses on smoother real-world performance, better obstacle awareness, and precise end-of-trip routing, among other improvements.
Key FSD v14.2.2 improvements
As noted by Not a Tesla App, FSD v14.2.2 upgrades the vision encoder neural network with higher resolution features, enhancing detection of emergency vehicles, road obstacles, and human gestures. New Arrival Options let users select preferred drop-off styles, such as Parking Lot, Street, Driveway, Parking Garage, or Curbside, with the navigation pin automatically adjusting to the user’s ideal spot for precision.
Other additions include pulling over for emergency vehicles, real-time vision-based detours for blocked roads, improved gate and debris handling, and extreme Speed Profiles for customized driving styles. Reliability gains cover fault recovery, residue alerts on the windshield, and automatic narrow-field camera washing for new 2026 Model Y units.
FSD v14.2.2 also boosts unprotected turns, lane changes, cut-ins, and school bus scenarios, among other things. Tesla also noted that users’ FSD statistics will be saved under Controls > Autopilot, which should help drivers easily view how much they are using FSD in their daily drives.
Key FSD v14.2.2 release notes
Full Self-Driving (Supervised) v14.2.2 includes:
- Upgraded the neural network vision encoder, leveraging higher resolution features to further improve scenarios like handling emergency vehicles, obstacles on the road, and human gestures.
- Added Arrival Options for you to select where FSD should park: in a Parking Lot, on the Street, in a Driveway, in a Parking Garage, or at the Curbside.
- Added handling to pull over or yield for emergency vehicles (e.g. police cars, fire trucks, ambulances).
- Added navigation and routing into the vision-based neural network for real-time handling of blocked roads and detours.
- Added additional Speed Profile to further customize driving style preference.
- Improved handling for static and dynamic gates.
- Improved offsetting for road debris (e.g. tires, tree branches, boxes).
- Improve handling of several scenarios, including unprotected turns, lane changes, vehicle cut-ins, and school buses.
- Improved FSD’s ability to manage system faults and recover smoothly from degraded operation for enhanced reliability.
- Added alerting for residue build-up on interior windshield that may impact front camera visibility. If affected, visit Service for cleaning!
- Added automatic narrow field washing to provide rapid and efficient front camera self-cleaning, and optimize aerodynamics wash at higher vehicle speed.
- Camera visibility can lead to increased attention monitoring sensitivity.
Upcoming Improvements:
- Overall smoothness and sentience.
- Parking spot selection and parking quality.
News
Tesla is not sparing any expense in ensuring the Cybercab is safe
Images shared by the longtime watcher showed 16 Cybercab prototypes parked near Giga Texas’ dedicated crash test facility.
The Tesla Cybercab could very well be the safest taxi on the road when it is released and deployed for public use. This was, at least, hinted at by the intensive safety tests that Tesla seems to be putting the autonomous two-seater through at its Giga Texas crash test facility.
Intensive crash tests
As per recent images from longtime Giga Texas watcher and drone operator Joe Tegtmeyer, Tesla seems to be very busy crash testing Cybercab units. Images shared by the longtime watcher showed 16 Cybercab prototypes parked near Giga Texas’ dedicated crash test facility just before the holidays.
Tegtmeyer’s aerial photos showed the prototypes clustered outside the factory’s testing building. Some uncovered Cybercabs showed notable damage and one even had its airbags engaged. With Cybercab production expected to start in about 130 days, it appears that Tesla is very busy ensuring that its autonomous two-seater ends up becoming the safest taxi on public roads.
Prioritizing safety
With no human driver controls, the Cybercab demands exceptional active and passive safety systems to protect occupants in any scenario. Considering Tesla’s reputation, it is then understandable that the company seems to be sparing no expense in ensuring that the Cybercab is as safe as possible.
Tesla’s focus on safety was recently highlighted when the Cybertruck achieved a Top Safety Pick+ rating from the Insurance Institute for Highway Safety (IIHS). This was a notable victory for the Cybertruck as critics have long claimed that the vehicle will be one of, if not the, most unsafe truck on the road due to its appearance. The vehicle’s Top Safety Pick+ rating, if any, simply proved that Tesla never neglects to make its cars as safe as possible, and that definitely includes the Cybercab.
Elon Musk’s Grok AI to be used in U.S. War Department’s bespoke AI platform
Tesla FSD (Supervised) v14.2.2 starts rolling out
