South Africa has withdrawn its draft AI policy after it was found to cite fictitious academic sources likely generated by artificial intelligence.
Government has pulled its draft National AI Policy after it was found to have cited fake academic studies as authoritative sources – almost certainly invented by AI.
Minister of Communications and Digital Technologies Solly Malatsi announced the withdrawal of the draft policy document on Sunday (26 April), just weeks after it was gazetted for public comment.
It follows a News24 report last week on at least six fictitious sources cited in the document that undermines a policy intended to build trust in AI.
The draft policy outlines initiatives and guardrails that would allow use AI as an agent for “inclusive economic growth, job creation, cost reduction, and a developing Africa.”
‘Internal questions initiated’
“Following revelations that the Draft National Artificial Intelligence Policy published for public comment contains various fictitious sources in its reference list, we initiated internal questions which have now confirmed that this was the case,”
said Malatsi.
“This failure is not a mere technical issue but has compromised the integrity and credibility of the draft policy. As such, I am withdrawing the Draft National Artificial intelligence Policy."
“South Africans deserve better. The Department of Communications and Digital Technologies did not deliver on the standard that is acceptable for an institution entrusted with the role to lead South Africa‘s digital policy environment."
“The most plausible explanation is that AI-generated citations were included without proper verification. This should not have happened."
This “unacceptable lapse” proves why vigilant human oversight over the use of AI is critical,"
he added.
“It’s a lesson we take with humility. I want to reassure the country that we are treating this matter with the gravity it deserves. There will be consequence management for those responsible for drafting and quality assurance.”
The Department of Communications and Digital Technologies initially responded that the fictitious citations did not affect the policy’s substance, integrity or policy direction.
This is contradicted by the withdrawal notice over the weekend, suggesting a deeper look at the entire drafting process will now commence.
Not the first to be caught out
AI has been known to dream up fake answers, often with serious consequences.
In at least two cases in SA courts, lawyers have been rebuked for apparently fabricating legal citations.
The policy withdrawal is an embarrassing climbdown by the government, raising questions about other policies previously submitted to parliament for approval.
Public Interest SA says the episode raises a more troubling, systemic concern.
“If such a glaring lapse could pass through the Department’s internal processes, what confidence can the public reasonably have in other representations made by the Ministry?"
“South Africa cannot afford policymaking that is casual with facts, cavalier with process, and costly to those who engage in good faith – least of all when it concerns the urgent project of inclusive economic participation in the digital age.”
Public Interest SA has called for an independent audit of the document and its sources, as well as public disclosure of the drafting process, including all contributors and verification steps.
Fake research not the only issue
The draft policy has been criticised for reasons other than fake academic citations.
Prominent SA tech investor and former CEO of Google SA Stafford Masie points out that it intends to create no fewer than seven new institutional bodies, including an AI commission, AI ethics board, an ombud and a regulator – all before a single rand has been committed to AI.
He also criticised the policy direction of reskilling workers for the new AI economy: “The correct response is not to reskill people for jobs that may no longer exist, it is to ensure that South Africa is where AI companies build, train, deploy, and employ.
“That requires infrastructure, incentives, and speed. It does not require seven new government bodies,”
says Masie.
Ironically, the policy document advocates containment against high-risk AI outputs in areas such as credit scoring, law enforcement and healthcare.
Accountability?
Here’s what the policy says about accountability:
“Public-sector systems are to be held to higher standards of sufficient transparency and accountability. Institutions deploying AI must establish traceable lines of responsibility, with a named accountable official or entity.
“Oversight institutions, including the proposed AI Ethics Board and AI Regulatory Authority, will monitor compliance, conduct audits, and assess system fairness, including gender and human rights impact assessments,”
it adds.
“These institutions are mandated to evaluate AI’s societal effects, enforce algorithmic transparency, and ensure the legality and contestability of automated decisions.”
All of which suggests the next draft AI policy document – and future policy proposals of any kind – will be subject to the kind of scrutiny seldom seen before, along with more robust interrogation of authors and contributors.
Source: Moneyweb
