This is a hard one to answer because bad actors try to hide their actions and cover their tracks. Also, human agents with the same agenda as the bots might be in the mix, muddying the data.

Starting with some simple stuff: chat bots have been around since the mid 60s. Early ones were very primitive, but were obviously an important stepping stone to where we are today. Modern social media (as we think of it) took off around 2003, give or take.

So at the absolute earliest, it might be a reasonable guess to say "about 20 years".

But how widespread were the earliest bots? Probably not that widespread. As 20 years ago, things like hosted cloud platforms were not as omnipresent as they are today. So 20 years ago, most folks would have to run everything on their own hardware, or use compromised computers around the web (infected with malware), or maybe university computers to run their bots.

However, state-sponsored intelligence and agitprop type agencies have massive resources in some large developed nations. Such agencies are often early adopters of new technology. So that's some food for thought.