In the tech field, a 'cat and mouse game' is unfolding between developers and management regarding the use of AI coding tools. While 15% of companies have banned these tools, nearly all development teams (99%) continue to use them in secret, with only 29% of companies establishing governance mechanisms. This phenomenon highlights the challenges faced in managing AI tools, particularly the growing severity of security issues. 80% of respondents express concerns about potential threats, especially the 'hallucination' problem generated by AI. 47% of respondents are open to allowing AI to make unsupervised code changes.