End-User Privacy Concerns

Transcription recording: accurate production of searchable transcripts from meeting recordings
Gesture recognition: recognition and accurate interpretation of physical gestures
Room interpretation: the ability to count the people in a meeting room, compare this to the names on the roster, and compare this to the available seating in the room to help local and remote meeting participants while also maximizing meeting efficiency and resource utilization

The platform delivers security, privacy, and reliability in every aspect of the product suite. For example:

Facial recognition technology: On-screen and inmeeting, the output can be a basic name tag— or no display at all, if the user’s identity is needed solely as input to another function. The data behind the facial recognition algorithms is tokenized, meaning no face image data is transmitted to and from the cloud. When an individual records his or her face, this is a process called enrollment. The individual enrolls by contributing images of their face. During facial recognition in a meeting, the platform generates a number, which is then matched against the closest known identifier. Cisco uses a GPU on the local device to calculate a descriptor to reduce the risk of thieves and outside agents. Given that features like this are based on users’ identity, our design philosophy has been to reduce the risk of access of personal information at every step of the process.

Wake-word interaction: When Webex Assistant is activated using its wake word, “OK Webex,” it results in speech being streamed to the cloud-based speech engine. The wake-word service listens only for the wake-word pattern and does not stream until it is activated.

Data management: When an organization opts out or a user leaves an organization, all related data, including machine learning models, is deleted.