No. The commitment states “from the day of signature on“.
Of course it would be helpful to prepare old data sets and to release them as well (and some of us have started to do so), but it is not a requirement of the commitment.
I work with third-party data (such as SOEP) and signed a contract that I do not hand out the raw data to others. Does that violate point 1 of the commitment?
From our point of view, that would be a justified exception to point 1, as long as you provide that justification in the author note. Maybe the data holder has good arguments not to release the data, e.g. for anonymity reasons.
But point 12 says that we promote the values of open science at our institutions. That means, you could express your wish to the data holder that the data are released under an open license. If enough users of the data set do that, maybe they can achieve that change.
I typically work with interview data/ sensitive clinical data / ... that cannot be anonymized and published as Open Data. I cannot sign the commitment, as I would frequently violate point 1.
The commitment clearly states in the preamble: “to every guideline there can be justified exceptions”. If you can convincingly explain why your data cannot be published due to anonymity reasons, it is perfectly OK not to comply to point 1. If you sign the commitment, you simply should give that justification in the author note, or in the README file of your public project repository.
You should also think about making aggregate level data open. For example, if you analyzed your interviews by counting certain topics or keywords, you could probably publish these summary data without compromising the anonymity of your participants. This way, at least you analyses can be reproduced.
In any way, this would not be a reason not to sign the commitment.