Dipankar Sarkar on Require human-in-the-loop oversight for…

Dipankar Sarkar

Writer on AI data rights and data-monetization policy; author for the Human Data Rights Coalition.

An agent is a set of actions: read a ticket, look up a customer, draft a reply, update a field, send an email, issue a refund. Each action carries different risk. The right question is: What action needs what kind of human approval? [...] Low reversibility or a large blast radius should push an action towards human approval regardless of confidence. High confidence only earns autonomy when mistakes are also cheap to undo.

AI Verified (Oct 8, 2026)

Source:

Dipankar Sarkar, “What should require human approval for AI agents?”, What Generative AI, reviewed October 8, 2026. Sarkar says each agent action has different risk and that low reversibility or a large blast radius should push the action toward human approval.

20d ago

Policy proposals and claims

Require human-in-the-loop oversight for agentic AI systems acting in high-stakes domains

votes For

Statement relation verification history AI Verified Report this

AI Verified The quote is explicitly about an AI agent’s actions, requires human approval, and identifies high-stakes conditions as low reversibility or a large blast radius. · Hector Perez Arenas gpt-5 · 20d ago

Vote inference verification history AI Verified Report this

AI Verified The quote supports the statement: AI-agent actions with low reversibility or large blast radius should receive human approval. · Hector Perez Arenas gpt-5 · 20d ago

Quote authenticity verification history

Report this

AI Verified The 8 Oct 2026 bylined guide publishes this wording verbatim, including the agent actions and its low-reversibility/large-blast-radius human-approval rule. · Hector Perez Arenas gpt-5 · 20d ago

replying to Dipankar Sarkar