The phrase "let rip" reveals how far policy has moved

Politicians choose words carefully. "Let rip" implies a dangerous default state that requires active restraint. That framing puts the burden on developers to prove control, not on regulators to prove harm.

The biosecurity focus is strategic. Unlike abstract alignment debates, pathogen risk has existing legal frameworks, international treaties, and clear liability chains. It's the wedge issue that justifies intervention without solving all of AI governance at once.

Our principle — no model training on customer conversations without notice and consent — aligns with this direction. The same logic that rejects unrestricted model development applies to unrestricted data ingestion. Consent is the minimum viable control.

The lab's output will likely become a reference standard. Companies that already operate with explicit boundaries, disclosed third-party inference, and user-approved actions will face lower adaptation costs when those standards become mandatory.

— Edgar Melo Costa, founder

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