On 18 November 2026 the Competition Commission of South Africa announced that it had settled with X Corp. and concluded a voluntary undertaking with X.AI. Both companies had appealed the Media and Digital Platforms Market Inquiry (MDPMI) remedies to the Competition Tribunal. Both have now withdrawn those appeals. The statement is a Commission press release, so the terms below are the Commission's own account. We have not seen the underlying settlement documents.
What the Commission secured
The MDPMI's final report, released on 13 November 2025, required X Corp. to make its monetisation programmes available in South Africa. It also required workshops and onboarding help for publishers, plus digital literacy programmes for adults and children. Publisher support and literacy programmes were to run for three years, with premium account support limited to two.
The settlement replaces these with two-year commitments. They take effect once the Commission confirms the remedies have been amended. According to the statement, X Corp. will:
- launch a Monetisation Support Programme within six months of that effective date, with twice-yearly workshops on revenue tools, analytics and content strategy;
- provide X Premium for two years where required, after which publishers pay for continued access;
- record workshops and translate them into up to five vernacular languages, subject to sufficient demand;
- run workshops on how publishers can manage their content on X, including whether public posts are used to train Grok;
- provide digital literacy programmes.
X.AI LLC, which the Commission says is now known as SpaceX Super Intelligence LLC, gives a narrower undertaking. It will train South African news publishers on Grok user settings and robots.txt protocols. It will fund the materials, deliver the training twice a year starting within six months of the Commission withdrawing the original remedies, and keep it available for two years. Revenue calculations on X remain unchanged. Both deals cover only qualifying South African news organisations that are members of the Press Council or the Broadcasting Complaints Commission.
The case for the original remedies
The strongest argument for a harder line is that X was the outlier. The Commission criticised it for refusing to take part in the inquiry's hearings. Tech Policy Press reported that X was also absent from the final negotiation process and faced imposed remedies it might appeal. Google, by contrast, negotiated a package worth about R688 million over five years.
A regulator that lets the one non-cooperating platform shorten its obligations after appealing could encourage others to stay out and litigate. Publishers also have reason to worry that a training workshop is a weak substitute for a financial commitment. On that critique, the deal looks soft.
Why the settlement is still the better outcome
We think the critique overstates the loss. Three points stand out.
First, the remedies were never a cash transfer. They are information and access obligations: tools, workshops, translation and training on AI controls. Cutting the duration from three years to two weakens them modestly. It does not remove them. Appeals before the Tribunal could have left publishers waiting with nothing, and the Commission itself says its priority is "moving from remedies to implementation," in the words of MDPMI chairperson James Hodge.
Second, the X.AI part is a sound use of a voluntary undertaking. The original remedy would have required X.AI to train publishers on controls and opt-out protocols for the use of news content in training large language models. The undertaking delivers the same training on Grok settings and robots.txt, which is what a publisher needs to control AI scraping. Teaching publishers to use opt-out controls is a proportionate intervention. It leaves publishers to decide on their own terms whether to license, block or ignore AI use of their content, and it does not impose a licensing regime on model developers.
Third, the Commission's own statement shows the remedies were shaped by negotiation. It agreed to amend the X Corp. remedies and to withdraw those imposed on X.AI. That is a regulator trading scope for speed, a trade we would defend. Market inquiries under the Competition Act are meant to fix market features through targeted remedies. They are not meant to maximise the size of compliance orders.
What to watch
The design leaves open questions that the public statement does not answer:
- Effective date. The two-year clock starts only when the Commission confirms the amended remedies. Until then, "within six months" has no fixed calendar date.
- Eligibility. Publishers must attend at least one workshop, apply, and meet X's integrity criteria. X therefore keeps gatekeeping power over who accesses the monetisation tools.
- Reporting. X Corp. will report to the Commission, but the statement does not say whether that reporting will be public.
- Outcomes. Revenue calculations are unchanged. Nothing in the deal guarantees that qualifying publishers will earn more on X, so the Commission should publish uptake and revenue data after year one.
The broader lesson for African regulators is that the MDPMI has now produced a negotiated outcome with the most reluctant platform, without a Tribunal ruling on the merits. That is a useful template for jurisdictions with limited enforcement capacity. It would be more useful still if the follow-through is measured and published, so that proportionate remedies can be judged by whether they changed publishers' circumstances, not by the number of workshops held.
The settlement is not a structural fix, and the Commission has not claimed it is. It is a pragmatic, bounded set of obligations that platforms accepted rather than litigated. For a pro-innovation reading of competition policy, that is a better result than a long appeal over a three-year training schedule.