Skip links

How Telefonica Works With Responsible AI

As with most other telcos, Telefonica is looking for new business models, as their old model is becoming extinguished. Privacy and security is one area, where the Spanish telco is investing and seeing new opportunities. For over a year, it has worked with its own ethical AI principles.

Privacy, security, ethics, and responsible use of AI are all part of Telefonica’s ‘business principles‘ whose aim is to secure trust from clients, employees, suppliers, shareholders, and society in general.

The AI principles are being implemented based three pillars; governance, training and awareness and self-evaluation.

“You need a certain kind of culture in the organisation. Many employees do not understand why, for example, it is important to take into account human rights when developing a product or service. So this summer we launched our Ethics and AI course internally. Not only so that employees would know and understand how to apply our AI principles but also to provide them with specific tools on how to monitor a service’s potential impact,” explains Geert Paemen, Director of Sustainability and Non-Financial KPIs at Telefonica. “With the help of these tools we can check for example, to what extent the application of an algorithm is more likely to lead to a discriminatory result”

The AI principles of Telefonica cover following:

1) Fairness. We ensure that applications do not lead to biased results and unfair and discriminatory impacts.We guarantee that there are no discriminatory elements when AI learns and algorithms decide.

2) Transparent and explainable. We inform users of the data that we use and its purpose.We take sufficient measures to ensure the understanding of the AI’s decisions.We tell our users when they are interacting with an AI system.

3) Human-centric. We ensure that AI always respects human rights.We are committed to the UN Sustainable Development Goals. We contribute to preventing improper use of technology.

4) Privacy and Security by Design. To build Artificial Intelligence systems, we take special care of information security.We respect the right to privacy of people and their data

5) With Partners and Third Parties. We confirm the veracity of the logic and the data used by the suppliers.

Some big beautiful promises from Telefonica.

“We are rolling out a process by which for any new product or service being developed, the product manager must, through a self-evaluation, make visible and manage any negative or positive impacts the product or service could have on society or the environment; it is a way of incorporating an ethical evaluation into our value proposition process “by design” says Geert Paemen.

At present there is no external independent evaluation of this process, and Paemen says this is very difficult to incorporate ex-ante into a products design process.

“The important thing right now is to generate awareness amongst all business units that these kinds of considerations are important and are part of our company culture,” she says.

Anonymized Location Data

Through its Smart Step’ services, Telefonica sells anonymized location data. As long as the anonymization is done proberly, this is a good way of using data, and GDPR is not applicable when it comes to anonymized data.

“Anonymized data can be used for services that make an important contribution to society and the environment, for example through the design of more efficient transport systems or the tracking of the spreading of viruses, says Paemen, and adds:

“With respect to the anonymization process we incorporate certain safeguards into the service, so we can ensure privacy and security in every step of the way, and we collaborate closely with data protection agencies in all the countries we work in.”

Even with the toughest anonymization – done in collaboration with a regulator – there are risks of de-anonymization. Therefore the German Data Ethics Commission recently suggested to criminalize de-anonymization of personal data.

Embedded from the Beginning

The main challenge when working with ethics and AI, and technology in general for that matter, is that ethics and human rights considerations must be incorporated in business processes from the very beginning. 

“This means our teams need to incorporate different perspectives into their approach. Multidisciplinary teams, as well as training and awareness-raising are important and can definitely contribute to generating a responsible corporate culture as well as a clear top-down implementation of processes and standards,” says Geert Paemen.

Apart from working responsibly with data, which Telefonica believes is a competitive advantage, the telco is expanding into new sustainable business lines and also digital consultancy such as security services.