By Alex Lawrence

Late last year EY’s annual report on the top 10 risks for telecoms highlighted that the number one risk relates to security. As EY’s Tom Loozen commented at the time,

“For telecom operators, the challenges around cybersecurity are mounting and the rise of Generative Artificial Intelligence (GenAI) is putting data governance under strain. The latest EY CEO Outlook Pulse finds that four in five telcos agree AI is a force for good, but more than seven in 10 believe they must do more to mitigate against AI “bad actors” and pay greater attention to the ethical implications of AI.”

With this in mind, a research paper by Hideaki Takahashi at the University of Tokyo is timely. The paper introduces an open-source library called AIJack which can help developers to identify the security and privacy vulnerabilities in their AI models and systems.

While nothing is going to be exhaustive in such a fast-moving field, this is currently a set of more than 40 attack and defence algorithms that will enable users to test out how robust or otherwise their systems are.

Meanwhile, an evolutionary approach from 5G to 6G might be appealing from a commercial perspective, but 5G – especially in non-standalone – carries many of the same security risks inherited from previous generations. A team of Canadian researchers has highlighted the problems associated with the use of VPNs to prevent unauthorised access in 5G and, by extension, 6G unless something new is done. They have proposed “a more stringent zero-trust security architecture” based on what they describe as a Software-Defined Perimeter [SDP] which is more resistant to attacks such as man-in-the-middle and denial of service.

This is an intriguing article on its own merits, but it’s very notable that it proposes to solve problems in an evolutionary way, recognising that the basic principles of the Evolved Packet Core are likely to be passed on to whatever the 6G core ends up looking like.

Such an approach appears to reflect the emphasis by operators, whether individually at events such as last April’s 6GSymposium or collectively through organisations such as the NGMN, on taking a path beyond 5G that can be implemented incrementally and with clear benefits associated with each step.

The paper underlines that further research will be needed to account for some proposed 6G architecture developments such as elements of possible RAN-Core convergence.