I think it is possible that one of the first "brain hacking" technologies will actually be to do with the voicebox as this will likely be simpler than working out different peoples brain structures.
Everyones voicebox works essentially the same and when thinking words signals are sent to the voicebox that are the same as but not as strong as the sigtnals used to cause speech.
Nasa already have this working for astronauts for speechless communication in space. See here
I think technologies to be able to monitor human thought to be quite scarey and quite useful at the same time. A device that could be placed inside someones throat and self powered somehow may be a deffinate big brother style thing, but it would be great to see this used for good purposes such as monitoring what politicians think when creating laws in parliment.
It just depends on the application.

we would never expect the CIA to lie about leadership of countires that don't agree with USA Hegemony

Considering they are coming from an uninformed "I will believe the big company when it speaks" paradigm, you could come back with "Well, you may want to consider that Cisco Intrusion Detection Systems have been based on Linux for years and they have even started using Linux for the OS for thier Firewalls and new switches, as well as the Opensource Antivirus ClamAV as part of the Desktop security solution 'Cisco Security Agent'".... While the statements itself say nothing regarding the security of these products it certainly is attacking the mindset of the purchasing goons for your company with something they will relate to. Disclaimer: Yes I do work for Cisco.

I believe this is based on the Blue Pill attack (from the same person) which essentially is a hypervisor that mimics the underlying system to gain access to the encryption keys. The flaws in the attack are that it is complicated to fully mimic the underlying hardware in software, the main drawback being that the timings by the hardware would be out due to the software hypervisor layer and this may be detected by the underlying OS or software running underneath the hypervisor. However it may be possible to write a hypervisor that takes all things into account but this would be quite an extensive task. ie. it is quite complicated to do properly but fesible (from what I have read). Mimicing the underlying system and the software interface to this via a hypervisor would allow access to the encryption keys. The article says basically "this is first stage attack, will produce stage 2 when intel responds to this" so they obviously have not completed the extensive programming task to take all things into account. Intel have known about this issue for some time as I asked one of their lead engineers the question a few months back if Trusted Execution was known to be totally secure and he basically said that theoritically it could be broken and told me to google "blue pill".