Is open-source AI more dangerous than closed AI?

Closed AI can concentrate capabilities and risks behind less public scrutiny.

Closed AI concentrates power in a few unaccountable firms, while open-source enables scrutiny, security audits, and broad oversight that make risks easier to find and fix.

Open-source distribution allows malicious actors to permanently bypass safety filters and deploy harmful models without oversight.