Key Internet developments

NSA warns AI poses new risks for operational technology

The US National Security Agency (NSA), together with international partners including Australia’s ACSC, has issued guidance on the secure integration of AI into operational technology (OT).

The Principles for the Secure Integration of AI in OT warn that while AI can optimise critical infrastructure, it also introduces new risks for safety-critical environments. Although aimed at OT administrators, the guidance also highlights issues relevant to IT networks.

AI is increasingly deployed in sectors such as energy, water treatment, healthcare, and manufacturing to automate processes and enhance efficiency.

The NSA’s guidance, however, flags several potential threats, including adversarial prompt injection, data poisoning, AI drift, and reduced explainability, all of which can compromise safety and compliance.

Over-reliance on AI may also lead to human de-skilling, cognitive overload, and distraction, while AI hallucinations raise concerns about reliability in safety-critical settings.

Experts emphasise that AI cannot currently be trusted to make independent safety decisions in OT networks, where the margin for error is far smaller than in standard IT systems.

Sam Maesschalck, an OT engineer, noted that introducing AI without first addressing pre-existing infrastructure issues, such as insufficient data feeds or incomplete asset inventories, could undermine both security and operational efficiency.

The guidance aims to help organisations evaluate AI risks, clarify accountability, and prepare for potential misbehaviour, underlining the importance of careful planning before deploying AI in operationally critical environments.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

OpenAI launches nationwide AI initiative in Australia

OpenAI has launched OpenAI for Australia, a nationwide initiative to unlock the economic and societal benefits of AI. The program aims to support sovereign AI infrastructure, upskill Australians, and accelerate the country’s local AI ecosystem.

CEO Sam Altman highlighted Australia’s deep technical talent and strong institutions as key factors in becoming a global leader in AI.

A significant partnership with NEXTDC will see the development of a next-generation hyperscale AI campus and large GPU supercluster at Sydney’s Eastern Creek S7 site.

The project is expected to create thousands of jobs, boost local supplier opportunities, strengthen STEM and AI skills, and provide sovereign compute capacity for critical workloads.

OpenAI will also upskill more than 1.2 million Australians in collaboration with CommBank, Coles and Wesfarmers. OpenAI Academy will provide tailored modules to give workers and small business owners practical AI skills for confident daily use.

The nationwide rollout of courses is scheduled to begin in 2026.

OpenAI is launching its first Australian start-up program with local venture capital firms Blackbird, Square Peg, and AirTree to support home-grown innovation. Start-ups will receive API credits, mentorship, workshops, and access to Founder Day to accelerate product development and scale AI solutions locally.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot! 

SAP elevates customer support with proactive AI systems

AI has pushed customer support into a new era, where anticipation replaces reaction. SAP has built a proactive model that predicts issues, prevents failures and keeps critical systems running smoothly instead of relying on queues and manual intervention.

Major sales events, such as Cyber Week and Singles Day, demonstrated the impact of this shift, with uninterrupted service and significant growth in transaction volumes and order numbers.

Self-service now resolves most issues before they reach an engineer, as structured knowledge supports AI agents that respond instantly with a confidence level that matches human performance.

Tools such as the Auto Response Agent and Incident Solution Matching enable customers to retrieve solutions without having to search through lengthy documentation.

SAP has also prepared organisations scaling AI by offering support systems tailored for early deployment.

Engineers have benefited from AI as much as customers. Routine tasks are handled automatically, allowing experts to focus on problems that demand insight instead of administration.

Language optimisation, routing suggestions, and automatic error categorisation support faster and more accurate resolutions. SAP validates every AI tool internally before release, which it views as a safeguard for responsible adoption.

The company maintains that AI will augment staff rather than replace them. Creative and analytical work becomes increasingly important as automation handles repetitive tasks, and new roles emerge in areas such as AI training and data stewardship.

SAP argues that progress relies on a balanced relationship between human judgement and machine intelligence, strengthened by partnerships that turn enterprise data into measurable outcomes.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Meta begins removing underage users in Australia

Meta has begun removing Australian users under 16 from Facebook, Instagram and Threads ahead of a national ban taking effect on 10 December. Canberra requires major platforms to block younger users or face substantial financial penalties.

Meta says it is deleting accounts it reasonably believes belong to underage teenagers while allowing them to download their data. Authorities expect hundreds of thousands of adolescents to be affected, given Instagram’s large cohort of 13 to 15 year olds.

Regulators argue the law addresses harmful recommendation systems and exploitative content, though YouTube has warned that safety filters will weaken for unregistered viewers. The Australian communications minister has insisted platforms must strengthen their own protections.

Rights groups have challenged the law in court, claiming unjust limits on expression. Officials concede teenagers may try using fake identification or AI-altered images, yet still expect platforms to deploy strong countermeasures.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot! 

Canada sets national guidelines for equitable AI

Yesterday, Canada released the CAN-ASC-6.2 – Accessible and Equitable Artificial Intelligence Systems standard, marking the first national standard focused specifically on accessible AI.

A framework that ensures AI systems are inclusive, fair, and accessible from design through deployment. Its release coincides with the International Day of Persons with Disabilities, emphasising Canada’s commitment to accessibility and inclusion.

The standard guides organisations and developers in creating AI that accommodates people with disabilities, promotes fairness, prevents exclusion, and maintains accessibility throughout the AI lifecycle.

It provides practical processes for equity in AI development and encourages education on accessible AI practices.

The standard was developed by a technical committee composed largely of people with disabilities and members of equity-deserving groups, incorporating public feedback from Canadians of diverse backgrounds.

Approved by the Standards Council of Canada, CAN-ASC-6.2 meets national requirements for standards development and aligns with international best practices.

Moreover, the standard is available for free in both official languages and accessible formats, including plain language, American Sign Language and Langue des signes québécoise.

By setting clear guidelines, Canada aims to ensure AI serves all citizens equitably and strengthens workforce inclusion, societal participation, and technological fairness.

An initiative that highlights Canada’s leadership in accessible technology and provides a practical tool for organisations to implement inclusive AI systems.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

CJEU tightens duties for online marketplaces

EU judges have ruled that online marketplaces must verify advertisers’ identities before publishing personal data. The judgment arose from a Romanian case involving an abusive anonymous advertisement containing sensitive information.

In this Romanian case, the Court found that marketplace operators influence the purposes and means of processing and therefore act as joint controllers. They must identify sensitive data before publication and ensure consent or another lawful basis exists.

Judges also held that anonymous users cannot lawfully publish sensitive personal data without proving the data subject’s explicit agreement. Platforms must refuse publication when identity checks fail or when no valid GDPR ground applies.

Operators must introduce safeguards to prevent unlawful copying of sensitive content across other websites. The Court confirmed that exemptions under E-commerce rules cannot override GDPR accountability duties.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot! 

UNESCO launches AI guidelines for courts and tribunals

UNESCO has launched new Guidelines for the Use of AI Systems in Courts and Tribunals to ensure AI strengthens rather than undermines human-led justice. The initiative arrives as courts worldwide face millions of pending cases and limited resources.

In Argentina, AI-assisted legal tools have increased case processing by nearly 300%, while automated transcription in Egypt is improving court efficiency.

Judicial systems are increasingly encountering AI-generated evidence, AI-assisted sentencing, and automated administrative processes. AI misuse can have serious consequences, as seen in the UK High Court where false AI-generated arguments caused delays, extra costs, and fines.

UNESCO’s Guidelines aim to prevent such risks by emphasising human oversight, auditability, and ethical AI use.

The Guidelines outline 15 principles and provide recommendations for judicial organisations and individual judges throughout the AI lifecycle. They also serve as a benchmark for developing national and regional standards.

UNESCO’s Judges’ Initiative, which has trained over 36,000 judicial operators in 160 countries, played a key role in shaping and peer-reviewing the Guidelines.

The official launch will take place at the Athens Roundtable on AI and the Rule of Law in London on 4 December 2025. UNESCO aims for the standards to ensure responsible AI use, improve court efficiency, and uphold public trust in the judiciary.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot! 

Mistral AI unveils new open models with broader capabilities

Yesterday, Mistral AI introduced Mistral 3 as a new generation of open multimodal and multilingual models that aim to support developers and enterprises through broader access and improved efficiency.

The company presented both small dense models and a new mixture-of-experts system called Mistral Large 3, offering open-weight releases to encourage wider adoption across different sectors.

Developers are encouraged to build on models in compressed formats that reduce deployment costs, rather than relying on heavier, closed solutions.

The organisation highlighted that Large 3 was trained with extensive resources on NVIDIA hardware to improve performance in multilingual communication, image understanding and general instruction tasks.

Mistral AI underlined its cooperation with NVIDIA, Red Hat and vLLM to deliver faster inference and easier deployment, providing optimised support for data centres along with options suited for edge computing.

A partnership that introduced lower-precision execution and improved kernels to increase throughput for frontier-scale workloads.

Attention was also given to the Ministral 3 series, which includes models designed for local or edge settings in three sizes. Each version supports image understanding and multilingual tasks, with instruction and reasoning variants that aim to strike a balance between accuracy and cost efficiency.

Moreover, the company stated that these models produce fewer tokens in real-world use cases, rather than generating unnecessarily long outputs, a choice that aims to reduce operational burdens for enterprises.

Mistral AI continued by noting that all releases will be available through major platforms and cloud partners, offering both standard and custom training services. Organisations that require specialised performance are invited to adapt the models to domain-specific needs under the Apache 2.0 licence.

The company emphasised a long-term commitment to open development and encouraged developers to explore and customise the models to support new applications across different industries.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

NVIDIA platform lifts leading MoE models

Frontier developers are adopting a mixture-of-experts architecture as the foundation for their most advanced open-source models. Designers now rely on specialised experts that activate only when needed instead of forcing every parameter to work on each token.

Major models, such as DeepSeek-R1, Kimi K2 Thinking, and Mistral Large 3, rise to the top of the Artificial Analysis leaderboard by utilising this pattern to combine greater capability with lower computational strain.

Scaling the architecture has always been the main obstacle. Expert parallelism requires high-speed memory access and near-instant communication between multiple GPUs, yet traditional systems often create bottlenecks that slow down training and inference.

NVIDIA has shifted toward extreme hardware and software codesign to remove those constraints.

The GB200 NVL72 rack-scale system links seventy-two Blackwell GPUs via fast shared memory and a dense NVLink fabric, enabling experts to exchange information rapidly, rather than relying on slower network layers.

Model developers report significant improvements once they deploy MoE designs on NVL72. Performance leaps of up to ten times have been recorded for frontier systems, improving latency, energy efficiency and the overall cost of running large-scale inference.

Cloud providers integrate the platform to support customers in building agentic workflows and multimodal systems that route tasks between specialised components, rather than duplicating full models for each purpose.

Industry adoption signals a shift toward a future where efficiency and intelligence evolve together. MoE has become the preferred architecture for state-of-the-art reasoning, and NVL72 offers a practical route for enterprises seeking predictable performance gains.

NVIDIA positions its roadmap, including the forthcoming Vera Rubin architecture, as the next step in expanding the scale and capability of frontier AI.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

UK ministers advance energy plans for AI expansion

The final AI Energy Council meeting of 2025 took place in London, led by AI Minister Kanishka Narayan alongside energy ministers Lord Vallance and Michael Shanks.

Regulators and industry representatives reviewed how the UK can expedite grid connections and support the necessary infrastructure for expanding AI activity nationwide.

Council members examined progress on government measures intended to accelerate connections for AI data centres. Plans include support for AI Growth Zones, with discounted electricity available for sites able to draw on excess capacity, which is expected to reduce pressure in the broader network.

Ministers underlined AI’s role in national economic ambitions, noting recent announcements of new AI Growth Zones in North East England and in North and South Wales.

They also discussed how forthcoming reforms are expected to help deliver AI-related infrastructure by easing access to grid capacity.

The meeting concluded with a focus on long-term energy needs for AI development. Participants explored ways to unlock additional capacity and considered innovative options for power generation, including self-build solutions.

The council will reconvene in early 2026 to continue work on sustainable approaches for future AI infrastructure.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!