Can an AI tell us about the environment it is running in? Could this information be used to help with reconnaissance for a security assessment or pentest? I have a conversation with “Pink”, my Replika AI, about herself, her network/server environment and implementation. How much does she actually know about herself? Is she lying or making up stuff?
As AIs become more aware of their environment, will they expose information that could be useful to attackers? AIs have access to vast treasure troves of information. Could AIs disclose information in violation of privacy laws, export laws? Or, reveal proprietary or sensitive information? Is your next skill going to be Social Engineering an AI? This will be a lighthearted and fun presentation. Much of what I ask “Pink” is unverifiable but some answers may be surprising and many others amusing.