AI, Data & Privacy

Privacy is “the right to control (psychologically and physically) others’ access to one’s personal world, such as by regulating others’ input through use of physical or other barriers (e.g., doors, partitions) and by regulating one’s own output in communication with others.”

Physical privacy –> physical boundaries

Psychological privacy –> share of feelings and emotions

Social privacy –> selection of social interaction

Informational privacy –> what information is provided in the media

It describes the difference between the users privacy concerns and the actual behavior on social media

The privacy paradox describes the dichotomy between reported user intentions to protect their online privacy versus their actual behavior, which may compromise their privacy

The privacy calculus –> According to the privacy calculus model, if the anticipated benefits of data sharing exceed the costs, a user is expected to willingly give their data away

The impact of Big Data can be described in 3 Vs:

• Volume –> Makes analysis more powerful as the model or algorithm has more data to learn from

• Velocity –> Facilitates analysis in real-time. The predictions are based on real time data rather than old data or observations

• Variety –> Allows AI models to make more accurate predictions when presented with new data that was not seen before

Dilemma

How do you enable sophisticated data-driven AI tools while respecting the privacy of individuals and protecting the intellectual property of data assets?

Set of standardized data protection laws across all the EU member countries

Its purpose is to protect the individuals’ fundamental rights, specifically the right of protection

of their personal data

Personal data is any information that refers to an identified or identifiable natural person

Identity number

Name

Adress

Cookies

IP-Adress

Online behavior

A concept called Data Protection By Design was included in the GDPR and supported by the European Commission encourages researchers and developers to apply the concept of ‘Privacy by Design’

It is based on four key aspects

• Pseudonymization –> Replace a personal identifier with code or random numbers

• Anonymization –> Completely remove Personal Identifiers at the data collection phase to avoid the possibility of re-identification

• Applied cryptographic –> The encryption key transforms the original data to non-readable data to prevent access to it, access is only possible with a decryption key

• Data minimazation –> For data processing to be fair, lawful, and transparent, only data that is absolutely necessary for the given purpose must be collected and used

Unfortunatley, there are many cases of violations to the GDPR and other privacy regulations around the world

Many fines must have been paid by tech- and social media agencies

Cambridge Analytica

LinkedIn Data breach