Woofun AI reports that Anthropic CEO Dario Amodei testified before U.S. lawmakers that open-source AI development is following a "very dangerous path." Amodei explained that releasing powerful models openly strips developers of the ability to supervise usage, monitor abusive behavior, revoke access, or update security mechanisms. He noted that unlike closed systems, fully open models hinder ongoing safety governance, potentially leading to irreversible abuse risks.