On Linux, you don't download random stuff from the internet, e.g. a new browser. You get it from a central source, usually package manager, where it is verified and secure
Devs tend to make strong use of packages on GitHub, PyPi etc which have been targeted quite a bit with malware. Malicious snaps and
Linux software is written to need only as many permissions as needed, but not much more.
Hooboi. Depends on who writes the software. There are plenty of dumb devs for either OS, and I've had to yell at many for requiring their commercial software (built in Java with an X11/web front-end and exposed listening ports) run as root, usually because they didn't want to figure out the permissions needed to access a device. There's a surprisingly narrow intersection of devs who understand OS security and networking.
Linux is usually always updated because of the central update mechanism, so that vulnerabilities are fixed very quick
For OS packages, sure, but are all your Docker containers, snaps, flatpaks, and appimages updated whenever one of the underlying libraries had a significant vulnerability? How about that PPA, or the stuff you compiled from source a year ago?
Because people are increasingly using those for software not available on the base repositories
Linux users often have a false sense of security that leads them towards insecure practices, often for the same reasons as Windows users (I just want it to do X and work). While traditional signature-based antivirus doesn't help much for either OS, there are plenty of other controls to fill the space that most people/organizations can - but don't - implement on either OS.
On Linux, that includes strict management/review of software+code sources, SElinux/AppArmor enforcement, remote logging+review, and much more. These often conflict with Linux devs idea of "freedom" and thus area a hard sell.