The entire premise is that somehow hole punching is dangerous and unstoppable with webrtc. Enterprise grade NAT environments can easily deal with both STUN and TURN hole punching.
If leaking internal IP's from your network is a security risk, you need to reevaluate what you're calling "security".
None of the article's concerns are concerns to any competent sysamin
E: replaced 'exposing' to 'leaking' to avoid confusion
To clarify (because I misunderstood the meaning of the comment before clicking the link), the parent is referring to leaking the IP addresses themselves, not 'exposing' access to them.
