It depends on the terms of the deposit and for what security one is applying the rate to.
All costs must be taken into account, and a deposit account may be too slow to apply to derivatives.
For equities, the most honest net risk free rate should be applied. As an example, Peter Lynch has a rule of thumb that the only time one shouldn't be in equities is when their average yield is lower than Treasury yields.
For shorter term investments, one year or less, the best liquid net rate should be used as the risk free rate. This is because the valid application of the risk free rate for investment decisions should be the best risk free rate available because marginal opportunities can be more precisely screened. Liquidity is less of a concern for a slow moving investor, but if one's active, yet the money is tied up in liquidation then the cost of that risk free rate rises in terms of the opportunity cost.
If one is trying to price derivatives, the question becomes more complicated because a higher risk free rate implies a high priced option. If the rate is too high, implied volatility will veer too far from statistical volatility so most likely isn't the real risk free rate.
All in all, truly risk free assets do not exist, and their approximates are approximately equal in net yield, but there are times where a more precise selection can improve results.
For CAPM, the best risk free rate is the highest deemed risk free, so in this case, if deposits are outpaying Treasuries then it is surely the best, but it must also be cost adjusted.
The cost to have one's money potentially tied up in a bank collapse needs to be accounted for. From my understanding, this process even during the worst since the Great Depression in the United States paid quickly, so if Canada's is at least as fast as that, the rate can go unadjusted.
If these special accounts charge fees for early withdrawal then that must be taken into account as well.
CAPM assumes that a security with higher albeit correlated swings to the market is a superior investment. That correlated variance is applied to the market return less the risk free rate because a risk free rate is assumed to have no variance.
If it is indeed true that the risk free rate should be an alternative's higher rate instead, less return will be multiplied by beta, so marginal returners will need higher betas to remain competitive, according to the theory.
It should be noted that Fama & French's initial claims about beta were disowned after practitioners failed to beat the market. They currently claim that P/B and beta explain all excess returns. Needless to say, there hasn't been a long line of practitioners trying to win with that theory this time.