On 23 March 2017, his Majesty King Willem-Alexander visited us at The Hague Security Delta: a network in which businesses, knowledge institutions and work together on knowledge development and innovation in the field of security. The King was particularly interested in the Sweetie 2.0 project that is preventing child abuse. Together with the University of Leiden and Tilburg University we work for Terre des Hommes on the development of an advanced version of a ‘chat robot’. Using artificial intelligence, it prevents the abuse of children through webcam sex.
The Sweetie 2.0 software will be used for mapping, measuring and combating online child abuse. One of the most innovative functions is the automation of responsive communications with online chat partners through the use of a personified chat robot in various communication channels. This is a virtualized minor child (aka Sweetie) who will engage in a dialogue using a chatbot facility and 3D imagery, in which all the chats are analyzed by the underlying frame work and that recognizes ‘online predators’ by the way of rapprochement. The chat logs are stored and exchanged data are processed per chat to a profile for each chat partner. This profile can ultimately be used to identify repetitive patterns. University studies will be conducted during the first few months of the project, that investigate the best ways to respond to persons who make indecent proposals towards a decidedly minor.
Cyber Agent Technology
The Sweetie 2.0 solution has been implemented to enable the inspection of chatrooms using virtual identities that have access to different messaging channels. These identities can be operated manually with or without the help from autonomous software agents (chat bots). Agents can be scheduled to explore online chatrooms by autonomously interacting with real persons. Their conversation model can be customized, and plugins are available to communicate in most popular chatrooms. Investigators can collaboratively explore thousands of chatrooms at the same moment, 24 hours a day, 7 days a week in any time zone. The system will automatically detect anomalies and notify users that may decide to manually review recorded transcripts or use tools for statistical analysis.