Security Model
It gets harder the more secure you want to be.
It does not exist in isolation - asking how secure something is doesn't mean anything without defining what you want to be secure against. The tools you use cannot always protect you from mistakes you make, so they do not remove the need to be informed and careful.
No network can remain secure against a global adversary while still being easy and convenient to join automatically.
If it's convenient for you to connect to a network, it's even more convenient for global adversaries, because they have more resources.
This is not unique to Freenet.
Freenet's darknet mode is more secure against network-level attacks than opennet because it makes it hard for an adversary to connect to the network: they must convince a human to connect with them.
Connecting on darknet with people you do not know - or worse, automatically - defeats its purpose.