Anthropic co-founder: Open-source AI poses irreversible abuse risks and may lead to dangerous paths

According to Bitcoin News, Dario Amodei, co-founder and CEO of Anthropic, stated to lawmakers in the U.S. Congress that the development of open-source AI is entering a "very dangerous path." Once powerful AI models are released in open-source form, developers will lose effective oversight over the usage of the models, including the inability to monitor abuse, revoke access, or dynamically update security mechanisms, thereby increasing potential risks.

Dario Amodei mentioned that compared to closed model systems, fully open models are more difficult to implement continuous control over security governance, which may lead to irreversible abuse risks.