Yep, the devil is in the detail EoinMag and as usual, in the details they leave out.
I would think it's something like this. If you define a safe level as one which results in a 0% increase in cancer risk, then an increased chance of cancer from, say 1.0% to 1.00001% can be classified as unsafe consumption.
I notice that they do not tell you how much of an increase we are talking about; which is quite telling. In these circumstances they usually quote a percentage increase in your chances of getting whatever they are talking about.
You usually get something like "Increases your chances of contracting XXXX by 30%!" These figures are actually designed to make a small number seem large, by giving you a percentage increase over an unknown base chance. If instead they said something like "Increases your chances of contracting XXXX from 0.20% to 0.26%" people would be more likely to go "so what?", but it's the same increase.
In this case they don't even give us the percentage increase figure, so I smell an even more fragrant rat than normal.
Naturally, by focusing only on the increased individual risk factors to alcohol consumption and ignoring any countervailing decrease to other risk factors, they can make alcohol seem all bad.
They just have to ignore the longer life expectancy of consumers over abstainers. And they claim they care about helth.