Following yesterday’s Town Hall, as discussion was a bit high level and there was simply no time to get under the surface, some speakers or audience repeatedly mentioning why Chain Fusion was better than other blockchain interoperability mechanisms and tools (bridge & wrapped tokens) because of its (expected) lower risk to lose 
(bcz of external or internal hacks), I had to look it up while going deep.
I could find arguments that would justify why theoretically this would stand – relatively homogeneous infrastructure with high-end hardware requirements & seamless software upgrades across the whole network demand-- but they lacked mentioning the initial assumption: that nodes operators and devs on ICP network need to trust the NNS implement the ICP protocol in a as it should and operate transactions as expected by the users, in a trustless and censorship resistant fashion.
This is not quite assured given NNS role & massively unilateral decision-making power on the whole network.
Compared to other blockchain layers, ICP’s NNS gives it superpowers… While being its weakest link governance wise (ie “trust issues”).
Such valid arguments by developers/techies experienced in other consensus mechanisms or protocols owe to be addressed IMHO.
Though difficult to address in a 2’ chat on X I completely understand 
PS. I wrote theoretically because as it seems
(More here 
Question about ICP's subnet design - #23 by Kurtis0108)
subnet’ canisters operated by Dfinity foundation have higher security levels than other subnets.