Tech Diplomacy Newsletter 8-26 - Tech Diplomacy Network %

Tech Diplomacy News: progress and setbacks in AI, cybersecurity, tech regulations, and more

Global

More than 600 youth gathered in New York City for the Youth and AI Summit on International Youth Day this month, presenting a Youth-led Declaration on AI Policy and Governancemeant to help shape the UN’s Global Digital Compact. Assistant Secretary-General for Youth Affairs Felipe Paullier stressed that “meaningful youth participation is not about being consulted in a tokenistic way at the end of a process. It’s about involving them at the beginning.”

Global AI governance discussions are focusing too heavily on flashy risks like election deepfakes while ignoring the “real danger” of AI embedded in election infrastructure itself, such as voter databases, biometric systems, and results software. For Rest of World, Lisa Poggiali and Samson Itodo argue elections deserve a dedicated track at the next Global Dialogue on AI Governance in 2027, since without functioning elections, citizens lose their main check on how AI is used across every other sector.

China’s November 2025 pause of sweeping rare earth export controls hasn’t stopped Beijing from imposing narrower restrictions, sanctioning ten American firms in June and fourteen EU entities in July. With the pause set to expire in November this year, China’s dominance persists, with control of “around 75 percent of global rare earth mining output and 85 percent of processing capacity.” While analysts call the recent measures largely symbolic, markets are already bracing for Beijing to restore broader controls once the truce lapses, Ariyazka Prasetyo writes for Modern Diplomacy.

Read the July global digital policy roundup from Tech Policy Press here.

North America

Meta will pay up to 16.7 billion USD to resolve litigation brought by a coalition of 47 states,accusing the company of designing Instagram and Facebook to keep minors compulsively engaged while gathering data on users under 13 without proper authorization. California’s Rob Bonta praised the outcome as delivering “real change” and “enforceable protections,” though Meta isn’t admitting fault. Notably, roughly 30% of the settlement money is contingent on TikTok and YouTube adopting comparable safeguards and matching Meta’s financial contribution, a condition Meta’s top lawyer framed as essential since kids “move fluidly across dozens of apps,” Mark Hallam reports for DW.

The Trump administration is weighing sweeping new semiconductor tariffs that could extend beyond chips to products like laptops and data center servers, and Commerce Secretary Howard Lutnick reportedly favors a system that ties tariff relief to companies’ investment in U.S. chip manufacturing. Experts and officials are pushing back, with a tech official who previously worked in Trump’s first admin calling it “the single dumbest way imaginable to pursue American dominance in AI,” and economist Sujai Shivakumar from the Center for Strategic and International Studies cautioning that tariffs alone won’t “produce more skilled technicians,” “shorten the permitting timelines,” or “expand reliable power and water infrastructure.” Tech industry lobbyists also warn the move could backfire, since Taiwan produces over 90% of the world’s most advanced chips and U.S. domestic capacity lags far behind, Ari Hawkins and Gabby Miller write for Politico.

For the New York Times, David E. Sanger, Dustin Volz, Ana Swanson, and Julian E. Barnes investigate the chaos and contradiction defining the Trump administration’s AI security policy, notably visible in the Pentagon’s dealings with Anthropic. Defense Secretary Pete Hegseth banned the company after it refused to drop limits on surveillance and autonomous weapons use, but the National Security Agency kept quietly running Anthropic’s Mythos model anyway, warning that dropping it would mean “unilateral disarmament.” This same whiplash shows up in export policy, the authors argue, as officials briefly blocked foreign access to Mythos before reversing without explanation and are repeatedly reworking rules on selling advanced Nvidia chips to China.

President Trump has declared a national emergency and signed an executive order banning the purchase, import, and installation of certain foreign-made electrical equipment used in the U.S. power grid this week, citing risks of sabotage and cyberattacks tied to entities linked to sanctioned or embargoed governments. Trump’s order pointed to the surging power demands of AI and advanced manufacturing as reasons the grid’s vulnerability now carries higher stakes, covering equipment like large transformers and industrial control systems, though local distribution infrastructure is exempt. The Energy Department has 120 days to draft formal rules implementing the ban, and Secretary Chris Wright must also assess already-installed foreign equipment for removal or isolation, with China’s roughly 80% share of global battery and solar inverter manufacturing making the move particularly consequential for that sector.

Canada’s escalating trade war with the U.S. is exposing how dependent the country’s AI ambitions are on foreign chips and computing infrastructure, especially as Nvidia warns of price hikes up to 15% on its chips. While rising economic nationalism might suggest Canadians would rally around building more domestic data centers, public support is far from guaranteed, with communities increasingly pushing back over concerns about electricity costs, water use, and whether the economic benefits will actually stay local rather than flow to foreign parent companies.

A recent Trump administration presidential memo authorizes private companies to actively fight cyber-enabled criminal groups, marking a shift from purely defensive corporate cybersecurity work toward government-sanctioned offensive operations. Although “under which authorities these industry actors would operate is still up for debate,” the policy targets ransomware gangs and cyber scam networks while explicitly excluding state-linked actors. However, this distinction is increasingly blurry given how criminal and state-sponsored hacking often overlap in countries like Russia and North Korea, creating risk that companies could unknowingly cross into forbidden territory amongst other concerns, Nikita Shah and Emily Harding investigate for the Center for Strategic and International Studies.

Africa

The Committee to Protect Journalists is calling on DRC authorities to restore mobile and internet service in South Kivu’s Fizi Territory, cut off since July 3rd amid intensifying conflict, with CPJ’s Angela Quintal warning the blackout is “depriving local populations of their right to be informed about the conflict and other issues” and violating constitutional and international communication rights. Local journalists also say they can no longer relay broadcasts or publish online, as soldiers reportedly seized Starlink satellite equipment from several community radio stations.

Elon Musk’s Starlink is “taking another step towards entering South Africa” after nearly four years of regulatory deadlock, with SpaceX now directly engaging the Independent Communications Authority of South Africa on how ownership and licensing rules apply to satellite operators. While the main obstacle has been Musk’s resistance to South Africa’s Broad-Based Black Economic Empowerment policy requirement for 30% local Black ownership designed to address inequalities stemming from apartheid, a December 2025 policy shift now lets foreign firms “use equity-equivalent investment programmes to meet empowerment objectives.” This may give the satellite internet company and controversial tech entrepreneur a path forward, Solomon Ekanem writes for Business Insider Africa.

Uganda-based investment firm Legacy Hills Investments met with investors at events in Milan and Bologna to pitch Uganda’s startup ecosystem as the country seeks more foreign funding to support job creation for its young population. The firm also announced plans to launch Uganda’s first institutional venture capital fund, drawing from pension schemes and family offices, aiming to give startups an alternative to commercial banks that often deem early-stage businesses too risky to lend to.

INTERPOL’s African Cyberthreat Assessment Report 2026 found that AI is now driving 55% of reported cybercrime across the continent, with cybercrime-related losses more than doubling since 2024 and being fueled largely by scams and automated social engineering. The report also flags growing use of AI-generated deepfakes for sextortion and synthetic identities that combine real and fabricated data to bypass biometric verification. The organization also pointed to progress, noting 17 African nations updated cybercrime laws in 2025, and that coordinated INTERPOL operations led to an increase in arrests and monetary recovery.

Madagascar’s Digital Development Minister Mahefa Andriamampiadana highlighted digital identity as one of the country’s biggest achievements at a recent mobile communications trade expo, pointing to biometric national ID rollout Prodigy alongside major telecom sector reforms meant to create a more enabling regulatory environment. He said Madagascar is now focusing on AI and mobile money, but stressed digital public infrastructure remains “the foundation on which meaningful digital services can be built and scaled,” Joyce Onyeagoro reports for TechAfrica News.

The U.S. International Development Finance Corporation has pledged 62.8 million USD to initiatives related to rare-earth minerals in Madagascar, Malawi, Angola, and South Africa as part of a broader push to reduce reliance on China’s dominance of the sector. The funding comes as China maintains deep influence across African mining through state-backed financing in countries like the DRC and Zambia, with analysts warning some rare-earth ventures may face commercial challenges.

Asia

This week, Taiwanese prosecutors charged nine people, including employees from Nvidia and Super Micro, for “illegally exporting ‘high-end AI servers’ to mainland China,” with 74 units successfully reaching the mainland through routes via Indonesia, Japan, and Hong Kong, and another shipment of 56 intercepted in Taiwan. Prosecutors are seeking the maximum five-year sentence for the alleged ringleader, an Nvidia manager accused of authorizing the chip releases, and say defendants used tactics like a shell company in Japan and fake websites to dodge export controls. Both Nvidia and Super Micro said they’re cooperating with authorities, the Associated Press reports from Taipei.

India’s Union Cabinet approved Semicon India 2.0 last month with a 1.275 trillion INR outlaycovering chip design, semiconductor equipment and materials, fabrication, advanced packaging, research and development, and skills. Separately, 12 semiconductor projects have been approved across six states, backed by more than 1.64 trillion INR in investment commitments.

State-backed cyberattacks from North Korea, China, and Russia rose 7.5% in the first half of 2026, according to a threat intelligence report from cybersecurity firm S2W, with North Korea remaining the most active actor at 99 incidents and disproportionately targeting South Korea, followed by the U.S. Pyongyang’s hackers increasingly leaned on generative AI, deepfakes, and fake job recruitment schemes to infiltrate cryptocurrency markets and software development ecosystems. Chinese state-sponsored activity declined 17.5% in volume, but analysts say Beijing-linked groups shifted toward quieter, longer-term espionage focused on telecommunications and expanding into Southeast Asia and the Middle East.

Illegal cryptocurrency mining is draining power grids and increasingly tangled up with organized crime across Southeast Asia, as investigators find growing overlap between illegal mining and the region’s cyber scam networks. U.S. and U.K. authorities sanctioned Cambodia’s Prince Group last October over forced-labor scam compounds that laundered proceeds through crypto, seizing billions in Bitcoin tied to the group’s chairman, and Thailand and Indonesia have seen similar crackdowns, with Thai investigators dismantling networks running thousands of stolen-power mining machines. Experts like telecommunications researcher Saaidal Razalli Azzuhri argue the fix isn’t banning blockchain technology but ensuring miners “pay the full economic cost of their electricity” instead of passing costs onto the public, David Hutt reports for DW.

Southeast Asian governments should focus AI regulation on the “three Ps: permanence, proliferation, and perniciousness”, Singapore-based legal scholar David Tan shares in an interview with The Diplomat’s Luke Hunt. He predicts organized crime syndicates will keep expanding their use of deepfakes and suggests public vigilance rather than expecting enforcement alone to solve the problem.

At the UN’s AI for Good summit in Geneva last month, China sent a large government and corporate delegation to pitch developing nations on open-source AI models as a cheaper, more accessible alternative to American systems as U.S. frontier lab CEOs and Trump administration officials were notably absent. Semafor’s J.D. Capelouto frames this as part of a broader Chinese strategy of “token diplomacy,” mirroring how Beijing previously used cheap solar panels and EVs to set global industry standards, now extending that playbook to AI by offering low-cost models and training to countries in Africa, Latin America, and elsewhere.

Europe

Google’s new 890 million Euro fine under the Digital Markets Act, covering self-preferencing in search results and restrictions on app developers steering users away from Google Play’s payment system, is fueling a wave of private damages claims from European tech companies. Comparison-shopping firms have already won or are waiting on major payouts based on Brussels’ earlier 2017 Google Shopping ruling, ranging from hundreds of millions to billions of Euros, and the new fine could open the door to fresh claims covering more recent years of alleged harm.

The European Commission’s legally mandated review of its 2030 Digital Decade Policy Programme has slipped past its June 30, 2026 deadline and won’t be published until 2027. The delay comes as the EU continues to fall short of its own digital targets, such as ICT specialist employment reaching only 10.5 million in 2025 against a 20 million goal, and basic digital skills among citizens rising to just 60%, below the 80% target for 2030. A 2025 Council conclusion “called the skills shortage a ‘vital’ issue for the EU,” and flagged the need for “‘particular’ attention in reducing the gender gap in digital skills,” Théophane Hartmann reports for Euractiv.

Finland’s Defence Tech Hub has grown to bring together over 100 companies and research organizations working on defense and dual-use technologies. The network includes leading firms like satellite radar company ICEYE and IQM Quantum Computers, alongside institutions such as the Technical Research Centre of Finland and the NATO Defence Innovation Accelerator for the North Atlantic, spanning sectors from quantum computing and AI to autonomous systems and drones. Member of Parliament Jarno Limnell framed the expansion as one that “supports Finland’s security, Europe’s technological sovereignty and our shared competitiveness,” and other officials cited strong international interest and demand that has exceeded initial expectations, Mohib Ur Rehman writes for Quantum Insider.

The EU’s AI Office began enforcing the AI Act early this month, alongside new transparency rules covering several areas, such as chatbots disclosing they are AI and the labelling of deepfakes. Over 180 organizations have already signed onto the related Code of Practice, and enforcement now covers general-purpose AI model providers, including stricter obligations for the most advanced systems deemed to pose “systemic risks,” such as those tied to cybersecurity threats and loss of human control, while also banning manipulative or rights-threatening AI practices outright. Rules for high-risk AI systems have been pushed back to December 2027, or 2028 for regulated products, though new bans on non-consensual sexual deepfakes and AI-generated child sexual abuse material take effect this December.

The U.K. and Ukraine finalized a defense technology pact this week, with Prime Minister Andy Burnham and President Volodymyr Zelenskyy signing an agreement in Kyiv that gives Britain first-in-the-world access to Ukraine’s Avengers AI Labs, a database built from thousands of frontline cameras and sensors tracking enemy equipment, drones, and troop movements. In exchange, Ukraine gains ties to the UK’s research institutions and tech sector, which officials describe as the third-largest AI hub globally, with the arrangement expected to eventually extend beyond defense into areas like transportation and infrastructure.

The newly proposed EU Chips Act 2.0 improves on its 2023 predecessor by shifting focus beyond fab construction toward chip design capabilities, but owning manufacturing facilities doesn’t guarantee resilience if a chip’s design is locked to one production site. Rather, true flexibility comes from “fab-independent” design teams that can shift production across multiple foundries, Bram De Muer argues for EE Times, recommending policymakers treat multi-foundry design capability as a resilience metric worth investing in.

Latin America

Brazil has launched a one billion BRL tender for a new AI supercomputer to be built as part of a broader 2.3 billion BRL package under its national AI plan, which envisions 23 billion BRL in total AI investment through 2028. The government hopes the new machine will rank among the world’s top 10 most powerful, adding to Brazil’s existing dominance in Latin American supercomputing. The announcement includes a Portuguese-language AI training facility in Rio de Janeiro built in partnership with Chinese firms Huawei and iFlytek, alongside a new algorithmic transparency research center at the Federal University of Minas Gerais.

The Stimson Center’s Initiative for Knowledge in the Americas Strategic Program has proposed a 350-450 million USD, five-year plan to build shared AI infrastructure across Latin America and the Caribbean, arguing the region currently has no collective access to advanced computing power or foundation models controlled by a handful of global tech giants. The blueprint rests on five pillars: a multi-country research network focused on areas like agriculture and health, a regional compute facility paired with a shared data commons, over 1,000 training fellowships to build local talent, harmonized regulatory frameworks, and a dedicated “deal desk” to negotiate partnerships with major tech firms on more favorable collective terms. Blueprint authors Juan Carlos Navarro and Francisco Sagasti argue that by pooling resources across the region’s 650 million people, Latin American countries can avoid the trap of technological dependency and instead build lasting AI capability, framing the fragmented status quo as far more costly than the proposed investment itself.

A new World Economic Forum report argues Latin America has a major strategic opportunity in critical minerals, even as global demand continues outpacing supply. The region has built strong upstream mining capacity but still lags significantly in processing and downstream manufacturing, limiting how much value it actually captures from its resources. The authors identify six “priority areas for action”: infrastructure, technology and innovation, financing, policy, social impact, and talent.

A ransomware attack disrupted Colombia’s Ministry of Justice earlier this month, degrading public-facing services tied to illicit-drug monitoring and legal processes just five days before the country’s presidential transition. The acting Justice Minister denied that data was stolen despite early reports suggesting a leak, saying files were encrypted rather than exfiltrated and that recovery was underway, while the incident followed a prior warning from Colombia’s national cybersecurity emergency response team about heightened ransomware activity targeting the country. The nation has also seen recent attacks on the national tax authority and oil company Ecopetrol, with experts pointing to a wider Latin American trend of automated, increasingly mature threat networks exploiting cloud vulnerabilities and third-party access.

UNESCO hosted an “Ethical Horizons of AI” discussion series at the Lima International Book Fair in Peru earlier this month, building on its 2021 global Recommendation on the Ethics of Artificial Intelligence and Latin America’s 2026-2027 regional ethics roadmap. Former Peruvian President Francisco Sagasti spoke on AI’s potential to advance human development, emphasizing that “regional cooperation will be key” for Latin America and the Caribbean to move forward together, while tech law expert Dilmar Villena stressed the need for algorithms and databases “contextualised and designed for our own reality” rather than simply importing foreign AI systems. The series drew on a 2025 UNESCO-Peru joint report assessing the country’s AI readiness across legal, social, economic, scientific, and technical dimensions, with participants agreeing that effective AI governance requires stronger institutions, technical talent, and coordination between sectors.

Oceania

Australian authorities filed charges against two young men from Western Australia over their alleged role in a hacking operation that hid malware inside a widely-used open-source code package, ultimately compromising over 1,000 organizations globally. Investigators say the scheme siphoned off half a million login credentials and roughly 300GB of data, with cleanup costs projected to run into nine figures. The pair face over a dozen charges combined, ranging from tampering with computer data to laundering criminal proceeds, and prosecutors are pursuing sentences that could reach two decades for the most serious counts.

Two subsea cables off Perth’s coast suffered worrying faults this month, prompting cable owner Subco to call for an urgent federal police investigation into the “concerning” incident. Security expert David Brewster noted that only 16 cables are facilitating 99% of the nation’s international communications, and warned that similar cuts elsewhere have often been traced to deliberate sabotage by state actors like Russia and China. He also flagged that these cables are increasingly owned by foreign tech giants like Google and Amazon, and that no single Australian agency is clearly responsible for protecting or responding to damage, ultimately leaving a gap in accountability, Rachel Stowers and Emma Wynne report for ABC Perth.

Australia passed legislation compelling major platforms like Google, Meta, LinkedIn, and TikTok’s parent ByteDance to negotiate compensation agreements with domestic news outlets, or else pay a percentage-based tax on their local ad earnings. Prime Minister Anthony Albanese touted the policy as a template other democratic nations might follow, while Meta has previously pushed back, framing the requirement as unjust and potentially in violation of trade commitments between the two countries. Given that Washington is already scrutinizing Canberra’s tech regulations, analysts think Australia could face retaliatory tariffs similar to threats leveled at European nations over comparable digital taxes.

Australia is funding a new regional program to give Papua New Guinea and other Pacific nations access to AI governance and risk-management toolkits, committing roughly 420,000 AUD through its Southeast Asia and Pacific Cyber Program. Announced at Pacific Cyber Week in Port Moresby, the resources will be developed by Standards Australia and are meant to help government officials assess AI’s opportunities and risks while promoting ethical adoption.

New Zealand’s venture-backed companies create “unusually high enterprise value relative to the venture capital invested,” but the country risks losing much of that value overseas as successful firms relocate their “economic centre of gravity” abroad, with Rod McNaughton arguing for The Conversation that innovation policy should focus on the local talent and networks that successful firms often leave behind.

Pacific Island nations, including Papua New Guinea, Tonga, Vanuatu, and Fiji, met in Solomon Islands earlier this month to set regional priorities through 2030 for improving seismic and tsunami monitoring, focusing on better data sharing and infrastructure interoperability across the Oceania Regional Seismic Network. Countries agreed to form a technical working group on data governance and standards, and pushed for improved technology and better coordination between meteorological and disaster management agencies.

New Zealand Prime Minister Christopher Luxon plans to introduce a bill banning children under 16 from social media, implementing fines that could amount to 10% of global revenue for platforms that fail to comply and requiring age-verification measures like digital ID checks. The move follows Australia’s similar under-16 ban introduced in December, though the bill’s passage is uncertain since coalition partner New Zealand First has already signaled opposition.

Read full newsletter