The DPO’s Role in Responsible AI

US Supreme Court just blew up EU-US Data Transfers

Tòa án Tối cao Mỹ hủy phán quyết Trump v. Slaughter, khiến FTC mất quyền độc lập, phá vỡ nền tảng pháp lý của EU-US Data Privacy Framework. Nhóm vận động quyền riêng tư noyb kêu gọi Ủy ban châu Âu hủy quyết định phù hợp và ngừng sử dụng dịch vụ đám mây Mỹ, trong khi các doanh nghiệp sử dụng SCCs/BCRs cũng bị ảnh hưởng do đánh giá tác động dựa trên cơ quan hành pháp Mỹ. noyb dự định kiện lên Tòa án Công lý EU (CJEU) trong 2-3 năm tới.

Lập trình viên nên đọc bài này vì quyết định của Tòa án Tối cao Mỹ phá hủy cơ sở pháp lý của Chương trình Bảo vệ Thông tin EU-Mỹ, ảnh hưởng trực tiếp đến các quy trình bảo mật dữ liệu trong các ứng dụng cloud và hệ thống chuyển dữ liệu quốc tế của các công ty.

The Download: a startup has a solution for AI’s groupthink problem

MIT Technology Review's daily newsletter covers several tech stories: an Australian startup called Springboards has built an LLM named Flint designed to produce more diverse, less predictable responses than mainstream models like ChatGPT or Claude. Other headlines include OpenAI proposing a 5% government stake deal with the Trump administration, Singapore seizing a mansion tied to Nvidia chip smuggling, Anthropic's Fable 5 returning online after a US export ban was lifted, Meta building a cloud infrastructure business, PlayStation ending disc releases by 2028, a low-cost Chinese AI model competing with US giants, and Google losing its €4.1 billion EU antitrust appeal.

Teaching AI to run with the turbines

Woodside Energy VP for Digital Andrew Melouney describes how the company has evolved its AI strategy over a decade, from traditional predictive analytics to agentic AI systems. Key initiatives include a 'Startup Advisor' copilot for LNG plant operators and a maintenance intelligence platform that correlates SAP records with time-series sensor data, targeting a 15% reduction in maintenance hours. The company now runs ~50 AI agents in production. Core lessons: build on governed, trusted data foundations; transition from isolated point solutions to enterprise-wide agentic systems; apply structured governance including an AI council; and follow a 'think big, prototype small, scale fast' philosophy. The long-term vision is an autonomous enterprise where interconnected agents deeply interact with core workflows.

The data sovereignty rules African and EU firms can’t ignore

Data sovereignty, residency, and localisation have become core architectural concerns for organisations operating across Africa and the EU. The post distinguishes between data residency (where data is physically stored), sovereignty (which laws govern it), and localisation (strict in-country requirements). It covers the EU's mature GDPR regime and Africa's fragmented but growing landscape of national data protection laws. Key risks of non-compliance include regulatory fines, operational disruption, vendor lock-in, and foreign government access. Practical architectural approaches include region-specific deployments, separated storage and processing layers, encryption with customer-managed keys, zero-trust access controls, and sovereignty-aware AI using federated learning. CIOs and CTOs are advised to map data flows, validate provider compliance, and run regular sovereignty impact assessments.

Pacemaker manufacturer Medtronic warns patients cybercrooks may have swiped health data

Medtronic, the medical device manufacturer known for pacemakers and insulin pumps, is notifying patients that their health data may have been stolen in a cyberattack attributed to the ShinyHunters threat group. The warning comes months after the initial breach, with the company disclosing what categories of patient information were exposed.

OpenAI has reportedly offered Washington a 5% stake worth $42.6bn

OpenAI has proposed giving the US government a 5% equity stake worth roughly $42.6bn at its current $852bn valuation, according to the Financial Times. The structure would involve a public wealth fund where OpenAI donates shares rather than sells them, with other AI companies like Anthropic, Google, and Meta potentially participating. The proposal competes with Senator Bernie Sanders' more aggressive American AI Sovereign Wealth Fund Act, which would impose a one-time 50% stock tax on large AI companies. Significant legal and structural questions remain unresolved, including how a private company transfers equity to the Treasury, who administers the fund, and how it interacts with OpenAI's existing nonprofit-controlled ownership structure.

UNICEF says children are adopting AI three times faster than adults

UNICEF analysis across 10 countries finds 20 million children have used AI tools, adopting them more than three times faster than adults. About 13 million use AI for schoolwork, while over 2 million turn to it for personal advice. UNICEF warns that child protection frameworks are not keeping pace, noting children have less power to challenge AI systems despite greater exposure. The agency calls on governments and the private sector to invest in research, stronger laws against AI-enabled exploitation, safety-by-design, AI literacy programs, and connectivity equity.