Someone might blame me as naive on this issue. But now the security is way more complicated and the problem is the more it gets complicated the harder to tight all holes. Kind of paradox.
Conventional security solution requires users to understand lots of security jargons and to make decisions so they are responsible all afterwards. In abstracted way, there is a s/w entity, which exposes lots of entry points that external s/w entities to access. Conventional secuirty approach might go over all public interfaces and make sure the external party has the proper access right. Usually beefed up security means more restricted and granular control on who gets what matrix. Does this solve the security problem? For this to make, there are lots of things we need to assume:
- We have to able to enumerate all public interfaces effectively and provide security check
- Policy setter has to have a really good understanding on security setting. Generally I feel like we are asking too much from end users.
- App developers have to be aware of security settings and they shouldn't allow any loophole in their end. This is hard and application development is getting more complicated and harder and expensive due to security consequences.
My day dreaming idea is:
- An intelligent agent equipped with machine learning capability. For every user action, it intercepts and analyzes the consequences beforehand. This agent can talk to each other or Mother agent to exchange information and to be smarter on fighting with viruses.
I guess there are some researchers doing this already and wonder how come I haven't heard anything yet.