안녕하세요,
Victoria from Techpoint here,
Here’s what I’ve got for you today:
DRC’s Internet shutdown passes one month
The DRC’s Internet shutdown in the east has now lasted more than a month, turning what might have been a temporary wartime measure into a much bigger digital-rights problem. The Committee to Protect Journalists (CPJ) said on August 17, 2026, that mobile networks and Internet services remain cut in Fizi Territory and the city of Baraka in South Kivu after the shutdown began on July 3. The authorities have not publicly explained the cutoff in detail, although local reporting has linked it to security concerns as fighting intensifies in the area.
And the impact goes far beyond people being unable to open WhatsApp or browse the web. Calls, SMS, mobile money, and online communications have all been disrupted, affecting businesses, families, humanitarian organisations, and journalists. Even more worrying, CPJ says soldiers have seized Starlink equipment belonging to several community radio stations and other organisations when they tried to restore connectivity. One journalist told CPJ that the military said the equipment would only be returned when the war ended. Some local radio stations have consequently had to suspend news bulletins because they can no longer receive or relay material from other media outlets.
The timing is important because this isn’t happening in a peaceful corner of the country. Fighting has intensified around the Minembwe highlands, including Kalingi, Mikenge and Point Zéro, involving government forces and Twirwaneho rebels allied with the M23/AFC. The shutdown therefore comes amid a genuine security crisis. But that’s also why it matters: cutting communication during a war can make it harder, rather than easier, to know what is happening. Journalists cannot verify claims, humanitarian groups struggle to coordinate, families cannot check on relatives, and rumours can fill the information vacuum. CPJ says the security situation should not be used as a pretext for blocking people’s access to information.
And, unfortunately, the DRC has done this before. When M23 forces advanced on Goma in late January 2025, authorities cut Internet access and restricted some social media platforms as the rebels took the city. Connectivity in Goma was reportedly disrupted for more than a week, while X and TikTok were also blocked in several major cities in February, with authorities saying they wanted to curb misinformation around the conflict. So there is already a pattern of connectivity restrictions appearing alongside major developments in the eastern conflict. The latest shutdown is different in scale and duration because Fizi and Baraka have now been offline for weeks.
And the DRC isn’t alone. Uganda imposed an Internet shutdown around its January 2026 election, while the Republic of Congo experienced a two-day shutdown beginning on March 15, election day. Gabon has also faced prolonged social-media restrictions amid unrest, while 2025 saw election-related connectivity disruptions in countries including Tanzania and Cameroon. The pattern is becoming hard to ignore: governments increasingly treat connectivity as something that can be switched off when politics, elections, or security situations become uncomfortable. And in a conflict zone like eastern DRC, where people already have limited access to reliable information, taking away the Internet for more than a month risks making an already dangerous situation even harder to see.
OpenAI launches ChatGPT for teens
OpenAI is giving ChatGPT a teen-specific experience, rather than simply treating younger users like smaller adults. Announced on August 18, 2026, ChatGPT for Teens is designed for users aged 13–17, with stronger safety protections, healthy-use features and additional controls for parents. If OpenAI’s systems estimate that someone is under 18, or the user says they are 13–17, the account is automatically placed into the teen experience.
The biggest change is what ChatGPT is allowed to do around sensitive topics. The teen experience has extra safeguards around things such as graphic violence, sexual or romantic roleplay, self-harm, risky viral challenges and content promoting extreme dieting or body-shaming. OpenAI is also adding features designed to encourage healthier use, including nudges to take breaks during extended sessions. The company says the aim isn’t simply to block teens from AI, but to make ChatGPT useful for learning, critical thinking and building skills while putting stronger guardrails around the experience.
For parents, this is where things get more interesting. OpenAI has already introduced parental controls that allow a parent to link their account with a teen’s account and manage settings such as memory and chat history. There are also narrowly designed notifications for situations where OpenAI believes a teen may be experiencing acute distress. The company says these controls are intended to give parents more visibility and control without giving them access to everything their child says in private conversations.
This has been building for a while. In September 2025, OpenAI said it was developing age prediction because it wanted ChatGPT to respond differently to teenagers and adults. In November 2025, it published its Teen Safety Blueprint, and in December 2025 it introduced its under-18 principles for model behaviour. Then, on January 20, 2026, OpenAI began rolling out age prediction, using signals such as account and usage patterns to estimate whether an account likely belongs to someone under 18.
The bigger story is that AI companies are now having to figure out what “age-appropriate AI” actually means. ChatGPT is already being used by young people for schoolwork, research and everyday questions, so simply keeping teenagers away from AI isn’t particularly realistic. OpenAI is instead trying to build a separate experience around them, but that raises its own questions about how accurately AI can determine someone’s age, how much control parents should have, and where the line between protecting teenagers and protecting their privacy sits. For OpenAI, today’s launch is less the end of the debate than the beginning of a much bigger one about how AI should grow up alongside its youngest users.
SARS wants to make VAT digital
South Africa is about to make VAT a lot more digital and potentially a lot harder for businesses to hide from. On August 17, 2026, the South African Revenue Service (SARS) published a consultation paper proposing a new Digital VAT Model, built around e-invoicing, an interoperability framework, and e-reporting. The idea is to move VAT administration away from paperwork and retrospective checks towards the near-real-time sharing of transaction data. SARS is asking businesses and other stakeholders to submit comments by 16 October 2026.
In simple terms, SARS wants tax compliance to happen more or less in the background of the systems businesses already use. Instead of waiting for companies to submit VAT information and then checking it later, the proposed system would allow transaction data to flow securely between businesses and SARS. The tax authority says this should make compliance easier for legitimate businesses while giving it much better visibility into transactions, making it easier to spot fraud and close the VAT compliance gap.
Why does this matter? VAT is one of the biggest sources of government revenue, and SARS is increasingly using technology to improve how it collects it. The new model is part of SARS Modernisation 3.0, which is essentially an attempt to build a tax system that can see what’s happening in the economy much closer to when it happens. For businesses, that could eventually mean less manual paperwork and faster processing. But it also means their invoicing and transaction systems will need to feed accurate information into the tax system.
SARS has been gradually tightening and digitising VAT administration for years. One recent change came into effect on April 1, 2025, when South Africa amended its VAT rules around electronic services supplied by foreign businesses, including changes affecting certain business-to-business transactions. Then, from April 1, 2026, the compulsory VAT registration threshold jumped from R1 million to R2.3 million, while the voluntary threshold rose to R120,000. So the latest proposal is another piece of a much bigger effort to reshape how VAT works in a more digital economy.
The important thing is that this isn’t a new VAT rate or a rule that businesses have to implement tomorrow. It’s a proposal, and SARS says the final design, technical standards, rollout sequence, costs, safeguards and implementation requirements will be shaped through consultation. But if it eventually becomes reality, South African businesses could be entering an era where sending an invoice and reporting VAT to SARS become much more closely connected. In other words, SARS isn’t just trying to collect more tax; it’s trying to make the tax system see the economy in real time.
In case you missed them
- MTN at 25: How much has Nigeria changed with it?
- CBN’s second sandbox cohort quietly opens a testing ground for permission-based data sharing
What I’m watching
- John Lennox: Will Technology Change What It Means to be Human?
- The Lapsed Atheist: Charles Murray Makes the Case for God
Opportunities
- Flutterwave is hiring for several roles. Apply here.
- Moniepoint is hiring for over 100 roles. Apply here.
- Follow Techpoint Africa’s WhatsApp channel to stay on top of the latest trends and news in the African tech space here.
Have a wonderful Wednesday!
Victoria Fakiya for Techpoint Africa
