Future of work

Google launches Gemini 3 Flash for scalable frontier AI

The US tech giant, Google, has unveiled Gemini 3 Flash, a new frontier AI model designed for developers who need high reasoning performance combined with speed and low cost.

Built on the multimodal and agentic foundations of Gemini 3 Pro, Gemini 3 Flash delivers faster responses at less than a quarter of the price, while surpassing Gemini 2.5 Pro across several major benchmarks.

The model is rolling out through the Gemini API, Google AI Studio, Vertex AI, Android Studio and other developer platforms, offering higher rate limits, batch processing and context caching that significantly reduce operational costs.

Gemini 3 Flash achieves frontier-level results on advanced reasoning benchmarks while remaining optimised for large-scale production workloads, reinforcing Google’s focus on efficiency alongside intelligence.

Early adopters are already deploying Gemini 3 Flash across coding, gaming, deepfake detection and legal document analysis, benefiting from improved agentic capabilities and near real-time multimodal reasoning.

By lowering cost barriers while expanding performance, Gemini 3 Flash enhances Google’s competitive position in the rapidly evolving AI model market. It broadens access to advanced AI systems for developers and enterprises.

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

UNDP and UNESCO support AI training for judiciary

UNESCO and UNDP have partnered to enhance judicial capacity on the ethical use of AI. A three-day Bangkok training, supported by the Thailand Institute of Justice, brought together 27 judges from 13 Asia-Pacific countries to discuss the impact of AI on justice and safeguards for fairness.

Expert sessions highlighted the global use of AI in court administration, research, and case management, emphasising opportunities and risks. Participants explored ways to use AI ethically while protecting human rights and judicial integrity, warning that unsupervised tools could increase bias and undermine public trust.

Trainers emphasised that AI must be implemented with careful attention to bias, transparency, and structural inequalities.

Judges reflected on the growing complexity of verifying evidence in the age of generative AI and deepfakes, and acknowledged that responsible AI can improve access to justice, support case reviews, and free time for substantive decision-making.

The initiative concluded with a consensus that AI adoption in courts should be guided by governance, transparency, and ongoing dialogue. The UNDP will continue to collaborate in advancing ethical, human rights-focused AI in regional judiciaries.

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

AI-generated ads face new disclosure rules in South Korea

South Korea will require advertisers to label AI-generated or AI-assisted advertising from early 2026, marking a shift in how the country governs AI in online commerce and consumer protection.

The measure responds to a sharp rise in deceptive ads using synthetic imagery and deepfakes, particularly in healthcare and financial promotions. Regulators say transparency at the point of content delivery is intended to reduce manipulation and restore consumer trust.

Authorities in South Korea acknowledge that mandatory labelling alone may not deter malicious actors, who can bypass rules through offshore hosting or rapidly changing content. Detection challenges and uneven enforcement capacity across platforms remain open concerns.

South Korea’s industry groups warn that the policy could have uneven economic effects within the country’s advertising ecosystem. Large platforms and agencies are expected to adapt quickly, while smaller firms may face higher compliance costs that slow experimentation with generative tools.

Policymakers argue the framework aligns with South Korea’s broader AI governance strategy, positioning the country between innovation-led and precautionary regulatory models as synthetic media becomes more widespread.

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

Salesforce expands Stockholm office to advance agentic AI

The San Francisco-based software company Salesforce has opened a significantly expanded office in Stockholm, reinforcing its long-term investment in Sweden and its broader Northern European strategy.

A new location that reflects the growing demand for AI-driven enterprise tools as regional businesses increasingly adopt agent-based technologies across their operations.

Located at Sveavägen 20, the Stockholm office is four times larger than Salesforce’s previous space and has been designed to support hybrid work, collaboration and innovation.

The opening event highlighted the focus of Salesforce on real estate as a strategic enabler for AI transformation, bringing together employees, partners, customers and community organisations.

A launch that also featured the Agentforce Sweden Nonprofit Hackathon, where Swedish charities presented AI agent solutions to improve efficiency and impact.

Majblomman received SEK 150,000 for an autonomous financial aid agent, underlining Salesforce’s ambition to position the Stockholm office as a regional hub for agentic enterprise development and responsible AI adoption.

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

Segment Anything adds audio as Meta unveils SAM Audio

Meta has introduced SAM Audio, a new AI model that uses intuitive prompts to isolate and segment sounds from complex audio recordings. The release extends the company’s Segment Anything collection beyond visuals into audio and video workflows.

SAM Audio allows users to separate sounds through text prompts, visual cues, or time-based selections. Creators can extract vocals or instruments, remove background noise, or isolate specific sound sources in recordings without specialised audio engineering tools.

Meta describes SAM Audio as a unified model designed around how people naturally think about sound. It supports combined text, visual, and time-based prompts, enabling flexible audio separation across music, podcasting, film, accessibility, and research.

Meta says the model achieves strong performance across diverse audio environments and is already being used internally to develop next-generation creative tools. The approach lowers technical barriers while expanding the range of possible audio editing applications.

SAM Audio is available through the Segment Anything Playground, where users can test the model with sample assets or upload their own files. Meta has also made the model available for download, signalling broader ambitions to make audio segmentation a core capability of its AI ecosystem.

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

EU approves €1.8 billion clean energy boost through Modernisation Fund

The European Commission and the European Investment Bank have approved €1.8 billion in new clean energy funding under the EU Modernisation Fund, supporting 45 projects across 12 member states.

Portugal receives funding for the first time after becoming eligible in 2024, while total support from the Fund since 2021 has now reached €20.7 billion across 294 investments.

Financed through revenues from the EU Emissions Trading System, the Fund targets high-impact projects that reduce greenhouse gas emissions in energy, industry and transport, while improving energy efficiency and strengthening energy security.

In 2025 alone, total disbursements reached €5.46 billion, with significant allocations directed to Czechia, Poland, Romania and Hungary, alongside support for Greece, Portugal and Slovenia.

All projects approved during 2025 focus on renewable electricity generation, energy storage, grid modernisation and efficiency upgrades in public infrastructure and industry.

The Modernisation Fund plays a central role in supporting national climate plans, reducing dependence on fossil fuel imports and advancing the EU’s Fit for 55 and REPowerEU objectives, with further investment proposals scheduled for early 2026.

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

New AI agent helps organise your day

Google Labs has introduced CC, an experimental AI productivity agent designed to help users stay organised and improve daily efficiency. The agent connects with Gmail, Google Calendar, Google Drive, and the wider web to gain an understanding of each user’s day.

Each morning, CC delivers a ‘Your Day Ahead’ briefing to users’ inboxes, summarising schedules, key tasks, and updates. It can draft emails, generate calendar links, and highlight next steps, making it easier to manage appointments, bills, and other responsibilities.

Users can interact with CC directly by replying to emails or sending custom requests, allowing the AI to learn personal preferences, store ideas, and remember tasks. The interactive approach helps the AI agent become more tailored to individual workflows over time.

CC is available in early access for Google consumer accounts aged 18 and over in the US and Canada, initially for Google AI Ultra and paid subscribers. Those interested can join the waitlist via the Google Labs website to gain early access.

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

Brazilian bank executive promotes Bitcoin for diversification

Itaú Asset Management partner Renato Eid has advised investors to consider allocating between 1% and 3% of their portfolios to Bitcoin. The recommendation, described as a measured approach, aims to strike a balance between diversification benefits and protection against currency weakness.

As head of beta strategies at Brazil’s largest private bank, Eid stressed the importance of a long-term perspective rather than attempting to time market cycles. Bitcoin, in his view, should function as a complementary asset rather than a central holding in a portfolio.

The guidance highlights explicitly Itaú’s BITI11 fund, a Brazilian-listed Bitcoin ETF that began trading on the B3 exchange in 2022 through a partnership with Galaxy Digital. The fund currently manages about $115.6 million and offers regulated exposure to Bitcoin for local investors.

Brazil’s currency volatility supports the case, with the real hitting record lows in December 2024 before partially recovering. Eid linked the strategy to Itaú Unibanco’s wider crypto expansion and increasing acceptance of crypto allocations among central global banks.

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

Local experts highlight AI’s potential for regional development

A panel of local business leaders, educators and policymakers in the Columbia Basin convened to discuss how AI can be harnessed to benefit the region’s economy, workforce and public services.

Participants highlighted AI’s potential to streamline government operations, enhance training programs and support small-business growth through data analysis and automation.

Speakers emphasised the importance of investing in workforce education and upskilling, so residents can capitalise on AI-related opportunities rather than being displaced by automation. Partnerships between local schools, employers and community organisations were cited as key to building ‘real-world readiness’ for AI integration across sectors such as agriculture, healthcare and manufacturing.

Panellists also emphasised ethical considerations and the need for community engagement in governance frameworks to ensure AI tools are adopted responsibly and equitably.

They argued that thoughtful regional planning can attract high-quality jobs and help the Columbia Basin carve out a competitive place in the broader digital economy.

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

AI shows promise in scientific research tasks

FrontierScience, a new benchmark from OpenAI, evaluates AI capabilities for expert-level scientific reasoning across physics, chemistry, and biology.

The benchmark measures Olympiad-style reasoning and real-world research tasks, showing how AI can aid complex scientific workflows. Generative AI models like GPT‑5 are now used for literature searches, complex proofs, and tasks that once took days or weeks.

The benchmark consists of two tracks: FrontierScience-Olympiad, with 100 questions created by international Olympiad medalists to assess constrained scientific reasoning, and FrontierScience-Research, with 60 multi-step research tasks developed by PhD scientists.

Initial evaluations show GPT‑5.2 scoring 77% on the Olympiad set and 25% on the Research set, outperforming other frontier models. The results show AI can support structured scientific reasoning but still struggles with open-ended problem solving and hypothesis generation.

FrontierScience also introduces a grading system tailored to each track. The Olympiad set uses short-answer verification, while the Research set employs a 10-point rubric assessing both final answers and intermediate reasoning steps.

Model-based grading allows for scalable evaluation of complex tasks, although human expert oversight remains ideal. Analyses reveal that AI models still make logic, calculation, and factual errors, particularly with niche scientific concepts.

While FrontierScience does not capture every aspect of scientific work, it provides a high-resolution snapshot of AI performance on difficult, expert-level problems. OpenAI plans to refine the benchmark, extend it to new domains, and combine it with real-world tests to track AI’s impact on scientific discovery.

The ultimate measure of success remains the novel insights and discoveries AI can help generate for the scientific community.

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