Digital responsibility updates from around the world

This week, we tracked 12 tech policy developments across 8 African countries. From AI and online safety to digital infrastructure and finance. Swipe for the highlights and subscribe for the full weekly briefing. #AfricaTechPolicy #TechPolicy #DigitalPolicy

https://lnkd.in/dzryXYYd The EU KIDS Act (COM(2026) 681 final) is set to reshape the unit economics of Big Tech. If passed in its current form, the regulation targets platforms built on high-friction engagement, algorithmic profiling, and autonomous youth onboarding. Primary Candidates for Regulatory & Financial Sanction Short-Form Video & Visual Platforms (e.g., TikTok, Instagram/Reels): Highly exposed to bans on "addictive-by-design" mechanics (infinite scroll, auto-play, daily streaks) and targeted recommender feeds for minors. Conversational & Companion AI Apps (e.g., Character.ai, Snap My AI): Direct targets of new restrictions on persistent AI memory, parasocial bonding, and unverified minor interactions. Freemium & Social Gaming Networks (e.g., Roblox, Epic Games/Fortnite): High liability surrounding dark patterns, underage microtransactions, and unverified age gating. The 4 Financial & Investment Impacts Revenue & Growth Compression: Forced bans on targeted ads to minors, reduced engagement hours, and restricted under-15 account creation directly erode Customer Lifetime Value (LTV). Escalated Debt & Legal Risk: Fines of up to 6% of global turnover—plus daily periodic penalties and private right-of-action litigation—introduce severe balance sheet tail-risks. ESG & SFDR Divestment Pressures: Increased risk of Social ("S") pillar downgrades, forcing Article 8/9 sustainability funds and institutional asset managers to re-screen or divest. Binary Market Access Risk: Non-compliance can trigger ISP domain blockades and mandatory App Store delisting across all 27 EU Member States. At THAX (Tech Harm Index), we quantify digital harm to help institutional investors, asset managers, and commercial banks direct capital away from platforms that exploit societal vulnerabilities and toward sustainable digital infrastructure. Measure the impact. Mitigate the harm. Move beyond short-term financial returns. #EUKIDSAct #TechHarmIndex #THAX #DigitalServicesAct #TechRegulation #SustainableFinance #SFDR #ESGInvesting #CapitalAllocation #TechValuations #RiskManagement #ChildOnlineSafety #SafetyByDesign #BigTech #DigitalHarm #TechPolicy

A few months ago, Canada introduced Bill C-34, the Safe Social Media Act 📱⚖️ It would create new rules for social media & AI chatbot platforms, including age limits, transparency requirements, and a new Digital Safety Commission to enforce it all. Want a version kids can actually understand? Check out the child-friendly resources below! 🔗 https://lnkd.in/dDJcJzRz

𝗔𝗴𝗲 𝗮𝘀𝘀𝘂𝗿𝗮𝗻𝗰𝗲 𝗶𝘀 𝗿𝗮𝗽𝗶𝗱𝗹𝘆 𝗲𝗺𝗲𝗿𝗴𝗶𝗻𝗴 𝗮𝘀 𝗮 𝗸𝗲𝘆 𝗿𝗲𝗴𝘂𝗹𝗮𝘁𝗼𝗿𝘆 𝗽𝗿𝗶𝗼𝗿𝗶𝘁𝘆 𝗶𝗻 𝗘𝘂𝗿𝗼𝗽𝗲 The European Commission's proposed EU KIDS Act is the latest example. The proposal would introduce a new framework for protecting minors online, including age assurance requirements, age-based access rules for certain digital services and safety-by-design obligations. Its scope extends well beyond social media and could affect video-sharing platforms, online games, app stores, AI companions and conversational chatbots. At the same time, the proposal raises important questions around privacy, proportionality, fundamental rights and the practical implementation of age-assurance measures. Although the legislative process is only just beginning, organisations may already wish to consider how the proposal could interact with their products, services and existing compliance frameworks. In our latest article, we explore the proposal's key elements, potential implications for digital service providers and the developments to watch as negotiations progress. Learn more here: https://lawand.tax/4ho21nD or get in touch with Stéphanie De Smedt, Pjotr Heemskerk, Anna Karklina or Kirill Ryabtsev. #EUKIDSAct #digitalregulation #technology #privacy #AI #onlinesafety #lawandtax

29,000+ US government websites are now searchable through one AI front door. Pakistan should think even bigger. FBR. SECP. Passports. Licences. Regulations. Government schemes. One secure conversational layer across government, with APIs that eventually allow AI agents to help citizens complete transactions, not just find information. The future of e-government is not 100 better websites. It is one intelligent interface.

Australia Urges Big Tech to Back Online Safety Rules Australia is seeking stronger cooperation from major technology companies as the government moves to tighten internet safety rules and develop new regulations for artificial intelligence.... https://lnkd.in/dgWMCutE

Australia Urges Big Tech to Back Online Safety Rules Australia is seeking stronger cooperation from major technology companies as the government moves to tighten internet safety rules and develop new regulations for artificial intelligence.... https://lnkd.in/dpdfXbBn

DeepSeek crashed after 8 a.m. Beijing Time today, as its website and the app showed that its server was temporarily unavailable, the conversation service was partially interrupted, and the performance of the website/API declined. The Chinese AI firm said it is investigating the problem and trying to identify the cause. The malfunction lasted for over an hour.

There is a meaningful new development today, September 18. China’s Cyberspace Administration of China published a new draft regulation on minors’ safe use of online services. It materially reinforces the July 15 AI anthropomorphic-interaction regime rather than merely repeating it. Most importantly, the draft explicitly prohibits providers from offering minors “virtual relatives, virtual companions and other virtual intimate-relationship services.” It also requires providers to identify minor users and automatically prevent them from accessing prohibited services. AI services capable of affecting minors’ cognition must operate through a dedicated minor mode. The implementation mechanics are getting much more concrete. Providers would have to develop actual minor-identification capability, potentially using identity credentials, national digital identity systems, one-time facial verification and behavioral signals. Devices must include anti-bypass controls, while app stores would have responsibility for age-rating and restricting installation of inappropriate applications. There is also a direct architectural requirement: providers may not deploy algorithmic models designed to induce emotional dependence, addiction or excessive spending among minors. Minor-mode implementations would have to be filed with provincial cyberspace authorities and could undergo annual verification. And the enforcement ceiling is substantially stronger than a mere design recommendation: violations of several core provisions could draw fines of RMB 100,000–1 million where illicit proceeds are below RMB 1 million, with serious cases potentially resulting in suspension, app/site closure or license revocation. The proposal remains a draft, with comments open through October 17, so these penalties are not yet operative under this instrument. This is important for the July AI-companion rules because China is now building the surrounding enforcement infrastructure: identify the user → determine age → constrain the permitted relationship → prevent bypass → constrain the algorithm → verify the control → hold the platform and distribution layer responsible. That is considerably more than content moderation. It is governance of whether the interaction is authorized to exist in the first place. CAC — September 18 draft regulation and full text⁠

Every new gTLD round is a snapshot of how companies expect the internet to look for the next decade. In a new article for CircleID, I compared the two we have so far. 📱 𝟮𝟬𝟭𝟮: 𝗮 𝘄𝗲𝗯 𝗼𝗳 𝗮𝗽𝗽𝘀 𝗮𝗻𝗱 𝗱𝗲𝘀𝘁𝗶𝗻𝗮𝘁𝗶𝗼𝗻𝘀 → About a third of the world was online, roughly 2.3 billion people → The smartphone was becoming the default device, and the app store the main door to services → Around 240 million domains, almost all in .com, the legacy gTLDs and ccTLDs → The program reflected it: companies bid for the shelf labels of a consumer web, the places people visit and the businesses they look for 🤖 𝟮𝟬𝟮𝟲: 𝗮𝗻 𝗶𝗻𝘁𝗲𝗿𝗻𝗲𝘁 𝗼𝗳 𝗮𝗴𝗲𝗻𝘁𝘀, 𝗔𝗣𝗜𝘀 𝗮𝗻𝗱 𝘁𝗿𝘂𝘀𝘁 → About 6 billion people online, 74% of the world → More than 400 million registered domains, nearly 53 million of them in new gTLDs → A growing share of answers comes from AI systems that read the web on our behalf → The program reflects that too: AI labs and platforms protect their product identities and claim the language of protocols, endpoints and credentials 🛡️ 𝗪𝗵𝗮𝘁 𝗰𝗵𝗮𝗻𝗴𝗲𝗱 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗱𝗼𝗺𝗮𝗶𝗻 𝗶𝘁𝘀𝗲𝗹𝗳 It's still an address, and it's becoming a trust signal that machines read before they cite a source, call an endpoint or complete a transaction. The namespace is now treated as regulated infrastructure. 🌍 𝗪𝗵𝗮𝘁 𝗵𝗮𝘀𝗻'𝘁 𝗰𝗵𝗮𝗻𝗴𝗲𝗱 𝗲𝗻𝗼𝘂𝗴𝗵 Most of the internet's growth since 2012 came from Asia, Africa and Latin America. The applications still point the other way. 💡 2012 was about giving people more places to go online. 2026 is about #domains that software can resolve, verify and rely on. 👇 The full comparison between the two rounds, and what it means for brands and registrars: link in the first comment.