This was revealed during a webinar held by the Inter-Parliamentary Union (IPU) on Wednesday morning, in which Portfolio Committee on Communications and Digital Technologies Chairperson, Khusela Sangoni-Diko, presented on what Parliament is doing when it comes to AI oversight.
The IPU’s webinar comes as more than 60 national parliaments have taken action on AI – from legislative reviews and oversight inquiries to dedicated committee structures and capacity-building programmes for MPs and staff.
“Yet approaches differ considerably across regions and institutional contexts. Most parliaments are still navigating how to organise their engagement with an issue that is technically complex, is evolving rapidly, and cuts across multiple sectoral mandates.
“This webinar is the first substantive opportunity for parliamentary exchange since the adoption of the Kuala Lumpur Declaration on Parliaments and responsible AI in November 2025.
“It draws on the IPU’s monthly tracker of parliamentary actions on AI policy and brings together parliamentarians and staff from different regions to share parliamentary experiences, lessons learned, and emerging good practices,” the IPU summary said.
Sangoni-Diko explained that AI is moving faster than most regulatory systems in the world, and that the central question for parliaments today is no longer whether AI will shape societies, but whether democratic institutions themselves will shape AI.
“The rapid evolution of technologies themselves means that the terrain is fluid, presenting challenges for governments particularly in policy-making environments that favour consistency.
“Today, an apex issue may be data sovereignty; tomorrow, it is systems bias. As one issue is resolved, another one emerges. To some extent, this fluidity then explains the wait-and-see approach favoured by some governments,” Sangoni-Diko said.
Tomorrow we engage with the Inter-Parliamentary Union (IPU) on parliamentary oversight over Artificial Intelligence. The IPU is the global organisation connecting 183 member parliaments worldwide. It is a dedicated platform for elected legislators to debate global issues, promote… pic.twitter.com/K6tIA8r5qP
— Khusela Diko
(@KhuselaS) May 19, 2026
She added that at best, lawmakers are said to have a foundational understanding, which points to a gap in AI governance that carries the risk of unintended consequences.
She said that AI governance cannot be approached as a technical issue, but that it is a societal issue that cuts across every area of governance and public policy. She also touched on the recent debacle of the draft national AI policy being withdrawn.
“Earlier this year, the government released a draft national AI policy for public comment. I’d like to issue a disclaimer that the policy itself has not been without controversy and had to be withdrawn.
“However, that experience highlighted an important reality. The governance of AI itself must embody the principles we seek to protect: transparency, accountability, consultation, and public trust,” Sangoni-Diko said.
“One of the key lessons from my experience is that parliamentary oversight on AI cannot wait for perfect regulation. Parliaments must engage even while policy frameworks are still evolving.”
She explained that Parliament is building capacity for MPs through foundational training.
“At the same time, led by our deputy speaker (Annelie Lotriet), Parliament is exploring the responsible use of AI within its own institutional processes. We’ve also developed an AI-supported tracking tool known as govtrack that assists committees in monitoring resolutions, executive commitments, and public participation processes.”
She stressed that AI governance must remain firmly people-centred.
“Parliament itself must continue building technical expertise and institutional readiness to undertake effective oversight in a rapidly evolving field. Ultimately, AI governance cannot simply be about regulating technology.
“It must be about strengthening human dignity, democratic accountability, and inclusive governments,” Sangoni-Diko said.
The webinar also heard from Ireland’s James Geoghegan, who said that the country has had the first dedicated standalone AI committee in Europe since November 2024.
“Typically, every single parliamentary committee in these Houses of Parliament, we have two houses, a lower House and an upper House, that mirror government departments, and that legislation that is flowing from a government department would go through that parliamentary committee.
“This is slightly different, and this is a cross-cutting committee. It’s a special committee that has been established initially for a period of 2.5 years,” Geoghegan said.
“And it was done to reflect as Honourable Sangoni highlighted the changes that are taking place in AI in such a rapid manner, and that there would be ongoing parliamentary scrutiny as the executive, as the government, develops its strategies.
“So effectively what that means is that our parliamentary committee has oversight and scrutiny across all government departments in respect of their use and their adoption of AI, but equally it is serving as a role to foster greater understanding amongst parliamentarians, but also greater understanding amongst the general public,” Geoghegan said.
Deputy Speaker of the National Assembly, Annelie Lotriet, who was also in attendance, explained that Parliament is implementing an AI Maturity Assessment Framework, and they have found that parliamentarians are working in silos.
She added that they are incorporating AI into how they conduct oversight.
“We have to look at things such as our oversight formats, where we do public hearings, we have submissions, and (look at) how we can utilise AI to make that process easier, and how we can collate all the data, because we work in the 1000s and 1000s of contributions that the public gives.”
Sangoni-Diko later concluded by saying that, similar to the podcast roundtable, they will also be holding one on AI.
