I am not here addressing anything specific to brain/machine interfaces, but only in general terms, there are few practical obstacles to brain/machine interface development, at least from the side of the developers. Current statements of professional ethics hold engineers responsible for the consequences of their creations, but this responsibility is hidden from individual engineers through a variety of mechanisms. For one thing, few projects are realized by single engineers, or even single teams of engineers, so they are able to remain ignorant of the full impact of their products. Also, policies and laws protecting whistleblowers and conscientious objectors to big industry have throughout the West been selectively unenforced, especially in the past 15 years, so the opportunities for engineers to affect research from a moral prerogative have been roadblocked by fascists. Moreover, the rise of 'incrementalism' as a go-to statement of the expectation of ethical obligations of engineers for their creations reflects the disconnect between individual engineers and the consequences of their designs, effectively removing any ethical concerns from affecting research tangents, so long as these tangents pursue established - and fruitful - paths of inquiry.
Looking at the paper that you have referenced, I see that ethical issues are not treated seriously, confirming the account given in answer to your concerns above. There is no education in ethics evidenced in the paper, and less interest beyond mounting a BMI to a person with a debilitating injury and/or coma. Millian utilitarianism is an effort at wealth redistribution and demands that experience and conscience guide policy decisions, and frankly has no direct place in any serious consideration of of BMI unless this is necessary to convince the fabulously rich that they would be better off opening opportunities to the disenfranchised poor than building robots to replace them. But, indeed, the opposite is the case, as any likely future involves BMIs taking the place of SSRIs in the management of the human stock until it can be manipulated to fit the trans/post-humanist box that near-term techno-corporate neo-feudalism sets out as our collective end.