Governments, standards bodies and the tech industry need to work together towards developing codes of conduct, standards and laws that improve security and protection in the new era of Artificial Intelligence. This was the main message from Huawei’s AI Cyber Security Forum held in Brussels today.
Huawei launched its new White Paper, Thinking Ahead About Security and Privacy Protection, at the forum, which was held in the company’s Cybersecurity Transparency Centre in the EU capital.
The paper outlines Huawei’s practices in and perspectives on AI security and privacy issues, and it proposes a shared responsibility model for their governance.
“AI is already impacting positively on the lives of millions of people in the EU,” said Bob Xie, Director of the Huawei Cyber Security Transparency Centre, presenting the report. “But while AI brings substantial opportunities and benefits, it also faces challenges in security and privacy protection as it becomes more ubiquitous. We in the ICT industry must foster the trust that will allow Europeans to benefit fully from AI technology,” he emphasised.
First such forum
This was Huawei’s first AI Cyber Security Forum in Brussels and was well attended by industry and opinion leaders, as well as representatives of the EU institutions. A number of distinguished speakers and panellists graced the event, including Dr Frank de Jonghe, partner and quantitative and analytics services leader at Ernst & Young, who also spoke on the need for stakeholders to collaborate in creating a trustworthy AI world.
Other speakers included Kevin Wang, senior privacy expert at Huawei, and Joerg Thomas, Huawei’s EU Data Protection Officer, who explained the Huawei’s privacy protection practices to the audience.
Panellists in a discussion after the speeches moderated by Dr Thorsten Jelinek, European director of the Taihe Institute, a Beijing think tank, included Mr Wang; Leuven University Professor Bart Preneel; Sophie Revol, an attorney and data protection specialist, also from Ernst & Young; and; David Mudd, global digital and connected product certification director at the British Standards Institution.
“We support European efforts to strengthen 5G cybersecurity and will make sure our products comply with stringent standard frameworks across the Union,” added Huawei’s Xie. “We support ethical regulation for artificial intelligence and will continue a constructive dialogue with European policymakers and regulators in this regard.”
Technological sovereignty – with data sovereignty at its core – will be a defining factor for the future of European industries in the AI era. The ability of European companies to manage data transmission, control how their data is stored, and analyse data will be the foundation of their competitiveness in the digital age, the forum heard. “Huawei see its purpose as contributing the technology that enables European businesses to maintain that sovereignty,” said Mr Wang.
Huawei has defined 7 Governance Objectives on AI security and privacy in the white paper:
System security and controllability: The system must meet the security requirements of robustness, stability, and adaptability and can provide security attestation.
Transparency and traceability: Logics (such as prediction, judgment, and automated actions) generated by the system must be explainable and transparent.
Privacy protection: Personal information and data must be protected and managed in compliance with the GDPR and other applicable laws.
Fairness: To mitigate unfairness caused by procedures, decisions made by AI should be able to be challenged and effective remedies sought.
Data management: The integrity, accuracy, availability, confidentiality, and comprehensiveness of data must be ensured.
Competence: A solution provider should specify the knowledge and skills required to securely and effectively run AI, and those who deploy and develop solutions should follow these knowledge and skills.
Deployment objective assurance: AI deployment and application must meet the requirements of lawfulness, fairness, security, privacy protection, and no misuse, as well as deployment purposes.