Since it is doing it again, and since the recall was performed, and since they believe that it was done in a satisfactory manner, you're probably on the hook for the repair at this point.
A call to your dealer might yield some help from them and/or GM, but that's not a given.
Deoxit is a good contact cleaner that you can use to clean the contacts on the plugs and the BCM, plugging them in and out a number of times to help burnish the surfaces, then apply some more dielectric grease before plugging them in the last time.
I had the recall done on my 2011 some time ago. There's a small square pad that's supposed to be permanently secured to the BCM case using a special glue (epoxy I think, but ...), certain wires removed from the factory harness, then the harness re-secured with a zip-tie to the pad. The pad in my car came loose in less than a week but I've never had my lights turn on or off at the wrong times, so for me this is a moot point.
The problem is with those vehicles who are having the issue again. It's unsafe no matter how you slice it. I've seen only a couple in my area with the lights on while driving on the freeway, only to turn off when they hit the brakes. The real fix that would most certainly be permanent would be to redesign the brake apply sensor (fancy-schmancy, overly complicated replacement for a simple switch) and either replace it with a switch or just add a switch to the system. Brake lights have been turned on and off for decades with simple switches and even if they failed, it was a simple job that anyone could do. Replacing the "sensor" can be done by anyone, but then it has to be taken to the dealership to have the BCM reprogrammed to recognize the new device. This design is poorly conceived because it's entirely possible to have designed it where the BCM would automatically recognize any changes to the "sensor", renewing what they refer to as "homing" each time the key is turned and/or the brake is applied. There are enough other "nanny" nuisance messages that adding another would be fairly easy to do and perfectly acceptable to any owner. It would let the driver know that they need to go through a simple re-learn process, which would be to place the car in Park, acknowledge the message, then step on the brakes a few times (or until the computer said it was happy) in order for the re-learn process to be completed in-situ. But I believe that someone saw an opportunity to make some money in the service department by driving (pun intended) people into the garage when just a little bit more consideration could have been used to make the car able to sort itself out. I know this, that for myself, the more that a device is able to take care of the owner instead of take advantage of them, the better I'll like it and the more I'll sing its praises, which will drive (pun intended again) more folks to their sales floor for more vehicles that are designed better than this.
"Sensor" in quotes above is to indicate that "it" is more like a pair of volume controls hooked up in reverse, one getting "louder" as the brakes are applied, the other getting "quieter". Then the BCM has to compare them, validate them, and then decide whether to turn the lights on or off. This particular design fails the test of time and, therefore, the test of good design.
I'll end my rant now, but I think you get my point.