What changed
Anthropic’s 8 October announcement says most changes clarify existing restrictions for increasingly independent AI systems. The update consolidates restrictions on deceptive campaigns and artificial activity, including fake accounts and concealed influence operations. It also clarifies requirements for high-risk health and finance uses and adds controls for autonomous physical actions.
The revised policy takes effect on 12 November; the announcement date is not the effective date.
The Australian business takeaway
Use the lead time to inventory where Claude connects to other systems. For each workflow, identify its owner, permitted actions and human approval points. Pay particular attention to customer-facing content and any integration that can affect people or physical equipment.
Read the full policy against your specific use case. This is a supplier-policy change, not a new Australian law or a finding that your current workflow breaches it.
Practical next step
Assign one person to review the policy and document any workflow changes before the effective date.