California should wield targeted court tools, not blanket AI kill switches

Anthropic’s recent meeting with Governor Newsom highlights a growing debate: should courts have the power to halt an entire AI model, or just the components that pose a clear danger? Their stance is clear – narrow, case‑by‑case orders are preferable to sweeping “kill switches.”

A blanket shutdown of a model can cripple legitimate uses that benefit businesses and consumers. By focusing on disabling a specific agent or cutting off a single customer’s access, regulators can neutralise the threat without stifling the broader technology ecosystem. This precision mirrors how we operate at ELP Ventures, where we only act on customer apps after explicit approval.

The message for AI firms is simple: build safeguards into the architecture now, so regulators can target the problem area without pulling the plug on the whole service. Transparent audit trails and modular designs make that feasible. Companies that ignore this risk not only face legal hurdles but also erode public trust.

In practice, a measured approach lets the industry advance responsibly, keeping innovation alive while addressing genuine public‑safety concerns. — Edgar Melo Costa, founder

Founder’s notes are opinion. They are not investment, legal or professional advice.