Malicious Silicon BMC & Service Processor Risk Mitigation

Some projects are asking what Collaborative Research is doing about Malicious Silicon BMC & Service Processor Risk Mitigation.

Our private systems use motherboards sourced from Sun Micrososystems, Hahnspring (HP) ASUS (Taiwan) and SuperMicro.

Our engineering folks have taken our Supermicro based systems offline for a physical check of Malicious Silicon. We don’t expect to find anything but if we do we will report what we find to UK and EU authorities in the normal way.

We are also checking other vendor SP’s and BMC’s but over the remainder of H2 2018.

All our Service Processors run on a discrete physical network with no Internet connection and no VLAN service is used to run the ‘Service Network’.

There are many ‘Service Networks’ to segregate system duties for our NAI systems and we monitor interconnection attempts at the network layer. We have never detected SP anomalies in our monitoring to date.

All Internet connections are minimised for control purposes so by way of assurance we don’t think we are leaking information other than that intended.

We are also validating commercial devices we have (e.g. HP smart printers etc.) to see if they are doing anything unexpected. Some network and server redesign has completed and will be actioned in Q1 2019 to further reduce risks from ‘commercial devices’.

To those who plant this type of interdiction; serves you right if you get dodgy data!

Note that we use an awful lot of simulation scenarios using artificially generated data for our R&D tests.

This data is simulated or based on predictive AI and is designed to stress our internal programmes using custom techniques. Such material should not be treated as fact.