You are indeed correct that there are so far no direct detections of gravitational waves, and no serious scientists are claiming (apart from excusable moments of being carried away) that the BICEP result constitutes such a direct detection. However, there is absolutely no evidence that gravitational waves don't exist, and plenty of indirect evidence that they do. (Plenty? Try "GR has passed every test so far with flying colours" on for size.)
As regards the current epistemic character of the BICEP data, here's one way to phrase it. We have a complex model for a physical system - the Big Bang - which relies on a huge corpus of science. Some of it has been tested, like the mechanics of general relativity, some of it hasn't yet, like the existence of some form of quantum gravity, and some of it is only a vague conjecture, like the inflaton field. This model makes a number of predictions, and occasionally we can actually test those predictions.
The BICEP data is exactly such a test, and the data comes out in roughly good agreement with the models, though of course there is still a fair bit of leeway on the parameters. As it happens, to explain the constraints that the data place on the model, one must include gravitational waves into the mix, or one must really bend over backwards and produce some really contrived-looking model of inflation. (In fact, take that last sentence with a large grain of salt - I'm not sure at all that such contrived models do exist.) In light of that what the data means is roughly that
- inflation is right, and gravitational waves exist, or
- some other effect is going on within inflation that mimics gravitational waves as far as B modes are concerned, or even that
- inflation is wrong, and there is after all some other model that explains all observations, which may not even need gravitational waves at all.
Now, I'm pretty sure that if you go on the arXiv these weeks you'll find plenty of papers exploring what other ways there could be to produce the observed B modes. It could also be that everything needs to be replaced, and if there's another model which truly fits all the data then believe me, cosmologists will love to hear of it, so bring it up if you have one. In the mean time, we try to make do with the models that we have and the data which we can measure, and we remain open to the possibility of a better model coming along. That's the way of science, more than your rather clumsy (if I may say so) attempts to define scientific rigour so far.
To put it shortly, the BICEP data simply say that the CMB does have primordial B modes. This can be extended into a solid estimate of how much energy was present in the tensor oscillations during inflation, which includes gravitational waves and more exotic phenomena. More generally, we regard it as confirmation of some aspects of the whole, complex model (which, again, contains more strictly-speaking untested science than just gravitational waves), and this makes us more confident that the model is generally correct. In particular, we hold that the detection of B modes in the CMB (and not of gravitational waves, as you suggest) constitutes good evidence of inflation. Finally, since the other ways to explain tensor modes within inflation are even more contrived, we take the data as indirect confirmation of the existence of gravitational waves, though of course we remain open to other possibilities and looking forward to experiments like Advanced LIGO which will hopefully tell for sure.
Note also, in particular, that this is the second independent indirect piece of evidence for the existence of gravitational waves. The Hulse-Taylor binary is the first, but to explain it one only needs a rather standard part of GR. This second experiment in no way relies on that evidence, and I really can't see how you think that.
It seems from the discussion in the comments that your mind is made up that scientific rigour has somehow been dropped, and it is not my intention to be drawn into a polemic with you. I post this answer in the hope that you'll accept that there are in fact subtleties which you have not yet considered, but that is in the end up to you. Your concern over the presentation of the issue to the general public is well grounded, and indeed any scientific development of this calibre tends to produce outreach articles with simplifications that are perhaps undue, and which do not reflect the current outlook of the scientific community. However, in such cases it is less helpful to simply decide that rigour has been dropped than it is to approach things with an open mind to learn what the outlook of the scientific community actually is. Just a thought...