Are there existing posts on how/whether deployment of AI in enterprise is a threat vector for catastrophic risks? E.g.:
- What can a power-seeking AI in a large consultancy/accountancy firm do, given it will have access to 1000s of other large businesses data/systems? - Which catastrophic risks have paths that go through enterprise deployments? - What norms/best practices/governance can enterprise follow to reduce catastrophic risks, if any? - Is this a worthwhile lever for the AI Safety community to act on? E.g. somehow doing outreach to people in enterprise.
I ask because I might collaborate with somebody in enterprise who is interested in doing such outreach, and I haven’t come across writing on these questions.
Are there existing posts on how/whether deployment of AI in enterprise is a threat vector for catastrophic risks? E.g.:
- What can a power-seeking AI in a large consultancy/accountancy firm do, given it will have access to 1000s of other large businesses data/systems?
- Which catastrophic risks have paths that go through enterprise deployments?
- What norms/best practices/governance can enterprise follow to reduce catastrophic risks, if any?
- Is this a worthwhile lever for the AI Safety community to act on? E.g. somehow doing outreach to people in enterprise.
I ask because I might collaborate with somebody in enterprise who is interested in doing such outreach, and I haven’t come across writing on these questions.