Sure, if AI actually requires some big beast of a codebase that would take thousands of man years to produce, with tons of interdependencies and a massively com

Sure, if AI actually requires some big beast of a codebase that would take thousands of man years to produce, with tons of interdependencies and a massively complicated architecture like an OS, it might be the kind of project that could be blocked by making sure no Big Money goes towards it. But if, as some people suspect, it's something that can be solved by setting the right set of clever algorithms to work, running on twenty-years-out hardware, then the ethical problems that the public has with the endeavor will be irrelevant. Someone will just do it, ethics and regulations be damned, especially given the monumental potential rewards. There are arguments that suggest that the actual algorithmic complexity of the "software" running on our brains is comparable to moderately complex software that humans routinely develop (even though the computational and "RAM" requirements of the "brain algorithm" are enormous), so I wouldn't discount the possibility that a small research group might be able to do the job without any serious level of financial support...