Skip to lesson
Exit
Governance, compliance & Cursor's control plane1 / 2

1 min lesson

The trust equation: blocker → ally

Apply the trust equation: a reviewer says 'we'll never know which code an agent wrote, so we can't trust it.' Name the fear and the converting control.

Step 1 of 2

Every security objection to AI coding reduces to three fears. Name the fear, then show the control that converts the blocker into an ally. This is the single most reusable frame in the whole governance day.

The fear (blocker)
Unreviewable - 'AI dumps huge opaque changes'
The control (ally)
Scoped, reviewable diffs through normal PR review
What you say
'Every change is a scoped diff a human reviews - same gate as any engineer.'
The fear (blocker)
Undisclosed - 'we won't know what's AI'
The control (ally)
Disclosure + AI-code tracking
What you say
'AI involvement is tracked and attributable, not hidden.'
The fear (blocker)
Unaccountable - 'no audit trail, agents run wild'
The control (ally)
Audit logs + early security involvement + SoDSeparation of Duties. Short for separation of duties: the rule that authoring, approving and deploying a change never sit with one person. Auditors use the acronym; the control is the same. Press Enter for the full definition.
What you say
'Bring security in early; every action is logged and the agent is never the approver.'
Why this works

Each fear is a specific, answerable claim, not a vibe. By converting 'unreviewable / undisclosed / unaccountable' into 'scoped / disclosed / audited,' you turn the security team from a gatekeeper into a co-designer of the rollout.

Invite security in early. The blocker isn't the tool; it's being surprised by the tool. Early involvement is itself a control.