Having a WiFi interface is not inherently worse than having an Ethernet interface, which nowadays we consider a given.
There are risks with any software that can be accessed through a network and, indeed, models will be eventually phased out, turning them into sitting ducks. Just have a look at all the perfectly functional gear out there based on Windows. I dare you to connect and old Agilent or LeCroy oscilloscope, based on Windows, to a more or less open university network. But more than a risk to the whole network it's a risk to the instrument itself. If the software gets corrupted you might end up with an expensive door stop.
There has been a lot of publicity about the risks of the Internet of Things, but the risk itself depends on the application. Right now the worst culprits, for example, are web accessible cameras that have become bots used in denial of service attacks. That is indeed a risk to your network because it can saturate your Internet connection. And of course a hacked camera can be used as an entry point to explore your whole network and look for other targets.
But web cameras and, in general, sensors, are generally conceived so that you will access them from a remote location. The user of such a device will often define a NAT mapping in order to be able to access a builtin web server from a remote location. And that is certainly a security risk. There have been serious incidents and there will be lots more.
However, is this a likely scenario with a device such an oscilloscope? I don't think so. Of course it can be convenient to monitor an experiment from home. But I don't think it will be such a frequent usage. So, the risk is not the same.
Thanks to the scarcity of IPv4 addresses most Internet connections are configured with NAT (network address translation) which turns the router into a de facto firewall. None of the devices you plug inside your network can be accessed from the outside unless you explicitly set up a mapping in your router. With IPv6 and no NAT this will change radically and routers will need an explicit firewall functionality. Fortunately IPv6 address spaces are so large, it will be impractical to scan address ranges in order to discover vulnerable devices. I've been using IPv6 at home for a year now and I have set up a packet capture for the parts of my address space I am not using (a /48 network). I am still to see a single packet, let alone a scan, directed to my network, while the IPv4 addresses get hit several times per hour.
So, is it worse with WiFi? I don't think so. Moreover, WiFi access points make it easier to segment your network than network switches. On most you can create several SSIDs and you donĀ“t need to bother with assigning ports one by one to different VLANs (which can be a bothersome chore in a network switch). Also you can configure many access points so that wireless users can't contact each other, reducing the probability of "propagation" of an "infection" from one device to another.
In short: what I would suggest to manufacturers is to provide a way to disable all forms of connectivity in the instruments. Not just WiFi, but Ethernet and/or USB. That will be useful in student labs where you want to minimize risks. Those "security" preferences should be protected by some sort of admin password, so that a user won't be able to use the network ports of an instrument without permission.