That's not exactly true. Honda required approval from Japanese regulators (MLIT) to put their L3+ car on the road (the first L3+ car in the world that consumers can at least lease, although it's only a limited fleet of 100 and lease-only).
Honda Global | November 11, 2020 Honda Receives Type Designation for Level 3 Automated Driving in Japan
The CA DMV made it clear Tesla must get their approval before putting any L3+ cars on the road. There might be some jurisdictions that allow it without any sort of formal approval however.
There was some discussion on this elsewhere, like what happens if Tesla allows FSD (a L3+ version) to activate in some jurisdiction that doesn't require approval, but then the owner drives it or sets it to drive into a jurisdiction that does. Like even if Tesla puts a software geofence that disallows it from activating there, or from setting a destination there, the owner can drive the car there manually. Then technically that L3+ car is still in that given jurisdiction without approval. This is not a concern for the other fleets being tested, but for consumer owned cars that is a concern.
Right now autopilot is not allowed in some countries. Though it is legal to have a car with autopilot in those countries, you just can't use it. I would assume it will be the same for a car with FSD in a jurisdiction that doesn't allow it.
This site has a lot of detail on FSD laws in the US
Which states allow self-driving cars? (2021 Update) | AutoInsurance.org
Oh yes he is saying that. But he also said that the cars sold in late 2016 had everything they needed for Level 5. He also said there would be coast to coast demo (NY to LA) by end of 2017. Elon says a lot of interesting things. I try not to take him at his word but rather attempt to translate what I think he means. Otherwise it would appear mr Elon had either a vivid imagination or trouble with the truth. I’m a glass half full type of guy hence i “translate” his statements.
I know some people think Elon is a huckster, but I don't see it. As a development engineer I've been guilty of the same thing (on a smaller scale). I see what is possible in the final product, but I underestimate the effort it's going to take to get there and it takes a lot longer than I thought. I'm better than average, I've seen some engineers miss their targets by years.
When I was at Boeing I worked in a lab that tested all the electronics that was under development for commercial aircraft. We built the simulators to test the hardware. I started a couple of years before the 777 was developed and was there through the early flight tests. Because the 777 had more electronics than any commercial aircraft to date and a lot of new technology for the industry, we had to develop a lot of new interfaces. The earlier aircraft used point to point communication like a USB cable. One end had the computer reading the data and the other had the device sending it.
The 777 had a new data bus that had been in development for over 10 years that was more like ethernet. A bunch of devices sending data on one wire with one or more computers on the same cable. Much more complex problem, but the R&D people had worked out the protocols to make it all happen. However now we had to make simulation hardware to talk to each of these components in as realistic a scenario as possible. At the same time the people making stuff destined for the aircraft had their own implementation problems to deal with.
The guy running the team developing the simulator interface thought it was going to be done in 6 months. It took more than 2 years. Just for one piece of the puzzle. As he got close to done, the software people could get their part together, but they had delays too because the software ended up taking longer than they predicted.
Tesla is attempting to create the equivalent system to an aircraft's autopilot capable of complete end to end operation without the need for a "pilot" in the loop. Something modern aircraft are pretty close to capable of now, in an environment with far fewer edge cases, but they still require two pilots by law in every country. Flying commercial aircraft today is an amazingly boring job.
The first primitive aircraft autopilots were introduced in WW II. They have been evolving for 70 years and it's still not legal for the plane to be flying with nobody at the controls. The first primitive autonomous driving systems for cars were cruise control which debuted in the 1970s. Possibly the late 60s. By the 80s they had electronic cruise control that could handle hills. Autosteering only came along in the last decade (on production cars).
90% of self driving conditions are quite easy to handle, but the edge cases can be like trying to get a bunch of balloons through a room full of razor blades. The last 10% is always the toughest part of the problem.
The engineers envisioning the system see how it will work when the problems are all worked out and vastly underestimate the effort it takes to get there. When it's the boss who is the visionary it's even worse because the engineers don't want to disappoint the boss with negative predictions and they pad their already optimistic estimates to be even more optimistic to please the boss.
Oh and I recall a few years ago Elon was making one of his FSD predictions and added a caveat that while he thought level 5 was going to be there by x date (long passed now), he said the regulatory issues may prevent people from using it when ready. I haven't heard him talk about regulatory hurdles since, but he's aware of them.
I've always been skeptical of the dream that the world would be full of autonomous taxis and people would rent out their cars when they weren't using them for ride sharing. I think level 3 is very doable and likely soon. We'll see how the regulators deal with it, but that's going to be on the road in the next year or two. Level 4 is going to be a bigger lift and we might get there, but we may never get to level 5.
The last year has made governments in functioning democracies very aware of employment issues. Even if level 5 was proven to be viable, I'm not sure many regulators would allow it. Many corporations will push for it, but it would lead to catastrophic unemployment that could collapse economies. In more than half of US states truck driver is the most common job. That doesn't account for those who are driving other vehicles. The gig economy is made up of people who are barely holding onto the bottom wrung of the economic ladder and quite a few of those people would be put out of work by autonomous driving.
Historically the US has had more of a sink or swim attitude about workers than many other developed countries, so the autonomous driving laws might be more open here than many other countries, at least at first. Many states might go back to requiring drivers in all vehicles even if autonomous to keep people employed. At that point, the economics of autonomous vehicles isn't as attractive. Additionally the young in the US are much more interested in something more akin to European democratic socialism and they will be pushing for better employment over automation.