KiteRa survey: 13.4％ hit rule or law risk from AI agent data

KiteRa, which provides internal-rules SaaS, ran the fourth volume of its GRC survey to look specifically at data governance. The finding that 13.4％ of respondents had brushed against rule or law violations, and 8.0％ saw damage to their company, points to where AI agents can go wrong: not in the model's reasoning, but in the documents it is allowed to read. Old internal rules and data whose latest version is unclear become the basis for everyday decisions and processing.

The survey also shows how far that data travels outside official channels. Shadow IT was reported by 44.9％, shadow AI by 38.7％ and shadow data by 36.2％. Roughly half of respondents, 49.7％, had entered or shared internal rules or data with AI tools their company had not approved, or whose approval status they did not know. Among employees at companies with 1,001 or more staff, 42.6％ reported doing so.

KiteRa's reading is that the accuracy and freshness of the information an AI references can affect not only routine judgment and processing but also policy and legal compliance and corporate risk. The survey also found that about 31.1％ of workplaces are using or verifying AI agents, while 36.1％ are not using or considering them and 27.9％ do not know.