This is the problem with slots, without the benefit of a clearly defined pay table and 'known odds' (such as exists with table games), and with RTP being reached over hundreds of thousands if not millions of spins, it's basically impossible to prove that something has changed unless the software provider drops a clanger as Playtech did here, or RTG did with Fruit Frenzy.
Even where the RTP is shown on the pay table or help file, we have to trust that it's correct because there's no way to verify it. (Hi-Lo Gambler is an unusual case in that it didn't take a huge number of plays to prove that something was amiss, and of course it's not a slot.)
Microgaming say they don't change their slots after they've been released, but we know for a fact they changed the behaviour of the Mega Moolah slots and then changed them back again -
https://www.casinomeister.com/forums/threads/thinking-somethings-changed-at-microgaming.26802/
There's nothing to stop them shaving 1% here and 2% there off the payouts of their slots, a virtually undetectable change that would net them a fortune as the weeks and months and years rolled by - and of course anyone who dares to say they think there's something amiss is immediately confined to the tin foil hat brigade dustbin.
I'm not picking on MG here, the above holds true for any software provider, and Playtech have certainly been caught in the act here. Whatever the details of how the free play versions of the slots ended up paying more than the real money versions (accident or deliberate), the fact is that Playtech were reducing the payouts of at least two of their games by stealth.
What both the Finsoft and now Playtech situations make abundantly clear is that online casinos are still pretty much just the Wild West of gambling, often with regulation that counts for the sum total of bugger all.