1/ AI is creating the conditions for either the best or worst period in human history.
Which one we get depends almost entirely on governance choices being made right now.
Mostly by people who don't understand the technology.
Thread ๐งต
-The problem with Agentic AI isn't capabilityโit's containment.
-A standard software sandbox won't stop an agent that can execute shell commands. The moment it breaks out of user-space, the host is compromised.
-Today we're introducing the architecture for FARADAY: a hardware-rooted, 5-level privilege isolation architecture for AI agents.
-It uses Xen hypervisor-level isolation to physically separate an agentโs reasoning from its execution environment. (Patents Pending)
-Paper:https://t.co/uf8muRBBZ6
-#AgenticAI #Cybersecurity #AISafety
1/ AI is creating the conditions for either the best or worst period in human history.
Which one we get depends almost entirely on governance choices being made right now.
Mostly by people who don't understand the technology.
Thread ๐งต
7/ The technology to close that gap exists. The architecture for real human control โ not symbolic control โ is a solvable engineering problem.
It's just not the problem most of the industry is working on.
That's what needs to change.