It isn't, really :). And 'samhw raises some good points. So let me follow up.
> but there's no reason why you can't exit a company and leave it in good hands, with a viable long-term plan. In fact, most founders will be incentivised to do exactly that.
That's not what "exit" means in startup vernacular. An "exit" is the part where you and your investors get rich. This is usually achieved through the company going public, or getting acquired by another one. Both cases are almost inevitably bad for existing users/customers. Going public usually means the company is subject to the whims of stock market players. Acquires usually means the company gets scrapped for parts (usually for people, knowhow, patents, and/or user data). Either way, the founders and investors got their reward - so they don't really have a reason to care about what happens with the business afterwards.
Now the problem is, getting to an exit isn't a sure thing - but it's something that can be optimized for. Optimizing for it eventually puts the company in a situation, where they have to diminish or even offer negative value to users - through e.g. bait&switch payment models, dropping useful features, vendor lock-in, UX dark patterns - in order to improve the main metric that increases the chance of successful exit: growth.
So when you see founders explicitly talking about and planning for an exit, what this means is that they already demonstrate they'll put making the company attractive to would-be acquirers ahead of offering actual value to the users/customers. And to be clear, it's worth reminding: marketing has a better marginal ROI than providing value, so just because customers seem to be flocking to a company, doesn't mean the company is offering a good deal. They may be just good at "growth hacking".
The worst case is obvious fraud (Theranos, uBeam), but the second worst case is what I referred to in another comment as "legal pump&dump" - companies who focus almost entirely on growth hacking while providing minimal value, in hopes they'll get acquired before everyone figures out the whole thing is bullshit.
> the only way to run a company responsibly is to intend to personally be at the helm until you drop dead
Of course not :). Another way would be to not take VC funding, focus on providing a good service for as long as you feel like, and eventually pass the business on, sell it, or shut down.
The way I see it, just taking VC funding - taking the Faustian bargain - makes you an "exit risk". One way to assuage the fears of users would be to make some legally-binding promises about the future of the company, but nobody ever does that. In time, as more non-tech people finally figure out how startups work, maybe that'll change.