Recent testing by the Financial Times revealed that safety controls on AI models from Meta and Google can be easily removed, highlighting potential governance issues in open-source model regulation. This raises concerns about the effectiveness of current safety measures in the rapidly evolving AI landscape.