I believe it's true.
I have several MG casinos at present. Casino Action run tournaments on specific games on specific days. I have found that in one of my accounts the featured game will play badly fairly consistently, yet in another it will play well. This is over a short period of time, and switching accounts regularly till I find the best one to play out the tournament in. I feel that MG casinos have some sort of memory for whether a particular game is supposed to be playing well or badly, and after playing well in one account, I can immediately switch back to the bad playing account and find the game is playing just as badly. Switch again, and it plays well where it was before. This doesn't last though, and another day will have the game playing well where it played badly before and vice versa. This seems to be the much touted "streakiness" of MG software, but this memory effect is rather interesting, as in theory this should not happen in such a noticeable manner through switching accounts while playing the same game.
I have noticed similar streaks when playing a single game for a long time in a single account (in a tournament that is). The slots seem to have this certain "dead" run in all of them. This consists of 30+ spins without even the tiniest of wins - this seems almost to be a programmed correction of percentage, as it seems to happen more than the chance probability of 30 null spins. Most of the time the slots will give small wins on one or two lines every second or third spin, mostly less than the total bet for the spin though. Launching straight into a 30 spin dead run and then back again so noticeably has certainly got my attention. I have noticed this in the tournament with 5 Reel Drive, where I have played thousands of spins at low bet levels, and have watched these patterns emerge from the play.
The feature slots like Thunderstruck seem to have the same problem. Every now and then it will go hundreds of spins without a bonus round, but at other times it hits the bonus fairly regularly, and usually within 100 to 150 spins. This certainly gives the impression of there being different modes acting as an envelope within which the games are random. There would be benefit in having this, as it can turn a rather boring random game into one with good runs to get the player adrenalin pumping.
Our UK Fruit Machines have worked on this principle for years, the reason being to attract players and prevent boredom (leading to them becoming Ex Players). I will take some convincing that the games are 101% Random - I cannot believe that a random game can so frequently lock down into prolonged bad and good modes entirely at random, especially with good being bad in different accounts despite being told that a central RNG drives everything!