Has Earth's climate ever changed? Sure.
[1]
There are many feedbacks within the climate system, some negative as you mention but some also positive. A well-known positive feedback is the greenhouse gas effect, which in the present age is caused by anthropogenic fossil fuel emissions. Increasing greenhouse gases (GHG) causes the atmosphere to trap more longwave radiation and increase the temperature. Negative feedbacks can occur in the climate, such as the "Planck" feedback which describes the T$^4$ dependence of radiation emission--so as the planet warms it emits an even larger amount of radiation.
Climate models already take into account many of these feedbacks, but one of the largest unknowns is the effects of clouds and aerosols on the climate. The distribution of phase of clouds may change as the climate changes, which could possibly counteract the warming (as made popular in the hotly-contested Iris hypothesis [2]) or further enhance the warming.
A forcing related to clouds that further complicates their feedback is the aerosols that they form on. Let's say the climate warms a lot and the planet goes into a major drought. That will cause more deserts which will increase the amount of dust in the air. That dust (for a given amount of water) could potentially suppress precipitation in the clouds, thereby extending their lifetime and increasing the amount of sunlight reflected. Without fully knowing the distribution and effects of aerosols (and their respective impacts on clouds), an accurate estimate of cloud feedbacks is difficult.
A runaway effect on Earth could technically be possible. You talk about feedbacks as a mechanism to prevent that, but some mechanisms are not easily reversible. Thinking about the ice coverage in Earth's polar regions, if all of that ice melts for a given increase in T, decreasing T does not necessarily allow that ice to grow back to the same coverage...and especially not at the same rate.
So a changing climate is a complex system with many different interrelated components that cannot simply be predicted by a $dT/dt = dGHG/dt$ relationship. There are numerous feedbacks that are all dependent not only on each other but also sometimes-irreversible processes.