Christine Gadsby, Vice President and Chief Security Advisor of BlackBerry Secure Communications, has warned of the danger looming for the Global South during an MWC session focused on global governance.
“We know Latin America, for example, was recently just named the most vulnerable place in the world as far as cyber security goes,” said Gadsby. “So you’ve got a couple of things happening. You’ve got the challenge of the cyber security piece, and then you’ve got the lack of frameworks regulating how vendors behave to protect data. So that’s kind of a bad combination.”
The problem doesn’t stem from the number of vendors, Gadsby explained, but “the lack of coordination of frameworks on holding those vendors to some sort of standard. And every standard in the world is different.”
The answer is to harmonise the frameworks, she argues. “The effort is there, but the harmony really is not and then what happens… is you have a lack of control and you have a lack of transparency, so you’ve got a lot of data flowing around and not a lot of visibility into what’s happening to it.”
All of which has one, chilling result: “Citizens’ data is not protected, government communications aren’t protected.”
BlackBerry’s approach to AI and ethics
When asked about BlackBerry’s approach to AI and ethics, Gadsby was quick to point out the opportunities from a youthful population. But it also presents a security challenge.
“75% of the population of Africa is under 35 and 75% of that population in Africa are using AI tools,” she said. “They’re using AI tools to build business rapidly, which, again, provides opportunity but security challenge. And you’re seeing Africa come out with a lot of great payment systems, and they’re really leading in a lot of ways, but that creates that fundamental challenge when you have the lack of cohesiveness that I mentioned.”
Then there’s the variety of languages, with Gadsby pointing out that over 3,000 languages are spread across the continent. This brings governance challenges. “Governance must be co-created in the language and locally where it comes from,” she said. “I’ve seen too many companies try and implement ethical AI practices that were baked in California or in Brussels for AI systems that aren’t going to run anywhere close to that.”
So what does localised governance mean in practice? “It means sitting with the people you know in Lima and asking them, you know, what are your values? What do you care about? What does fairness look like to you? Because fairness in Peru may not be fairness in the US or in Canada,” said Gadsby.
“From a security perspective, AI that discriminates is also AI that’s vulnerable,” she added. “And that actually opens it up to vulnerabilities of manipulation. It opens up legal liability and loss of trust, which for us is non negotiable.”
All of which means that AI ethical governance must be tested with diverse populations rather than in Europe or North America.
“If large enterprise vendors and companies are making AI systems, they need to make sure and be responsible to test them in the where they’re going to be deployed, maintaining human oversight, especially for high-stakes decision making… And that’s got to start with the CEO, and it has to run down.”
Enforcing the rules
Before leaving the stage, Gadsby had one final message for attendees: the importance of enforcement. “We can all sit up here and say we have all these great ideas on how to make all of these frameworks and rules and governance and things. But if it’s not enforceable, it doesn’t work.”
She said nations should “take a page out of the playbook from places like the EU – they dangle the all important dollar in front of theirs, right?” So they don’t just ask companies to “play by the rules” but back that up with fines and the threat of being blocked.
BlackBerry, for example, is held accountable by the EU. “We have a whole programme team that has to run all of our audit. It’s really tough to keep meeting these standards, because we follow standards that are enforced, right?”
Her final point was to create centres of excellence: to follow a “hub and spoke model”.
“BlackBerry has done that… in Malaysia for security research, to bring those [experts] together. And it’s working. They put 5,000 people through that programme. Vendors do need to support that.”