Capacity development

NotebookLM Improves User Experience With Streamed Answers

Google has updated its AI research assistant, NotebookLM, to display AI responses in real time rather than all at once. This change reduces perceived waiting time by 30 to 40%, according to Simon Tokumine, the product’s lead.

The update follows the tool’s move to using more advanced ‘thinking’ models, which, while more capable, can take longer to generate responses.

NotebookLM, which recently launched mobile apps for iOS and Android, was previously limited to the web. However, the mobile apps currently lack features like Mind Maps and content generation. Still, the new response streaming feature marks a meaningful improvement in user experience.

Additional announcements at Google I/O 2025 include more control over Audio Overviews, allowing users to select shorter or longer summaries based on their preferences. Google also confirmed that Video Overviews are in development, following the positive reception of the audio feature.

While many of these updates are incremental, they reflect Google’s continued investment in refining NotebookLM as a practical, AI-powered research assistant.

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

Oracle and OpenAI target AI leadership with massive chip project

Oracle has reportedly acquired around 400,000 Nvidia GB200 AI chips valued at approximately $40 billion for deployment at a data centre in Abilene, Texas.

The location will be the first site of the Stargate project—a $500 billion AI infrastructure initiative backed by OpenAI, Oracle, SoftBank, and Abu Dhabi’s MGX fund, which President Trump announced earlier this year.

Once completed, the Abilene facility is expected to provide up to 1.2 gigawatts of computing power, rivalling Elon Musk’s Colossus project in Memphis.

Although Oracle will operate from the site, the land is owned by AI infrastructure firm Cruso and US investment company Blue Owl Capital, which have collectively invested more than $15 billion through financing.

Oracle will lease the campus for 15 years, using the chips to offer computing power to OpenAI for training its next-generation AI models.

Previously dependent solely on Microsoft’s data centres, OpenAI faced bottlenecks due to limited capacity, prompting it to end the exclusivity agreement and look elsewhere.

While individual investors have committed funds, the Stargate project has not officially financed any facility yet. In parallel, OpenAI has announced Stargate UAE—a 5-gigawatt site in Abu Dhabi using over 2 million Nvidia chips, built in partnership with G42.

A surging demand for AI infrastructure has significantly boosted Nvidia’s market value, with the company reclaiming its top global ranking in late 2024.

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

Quandela presents Belenos, a powerful 12-qubit quantum computer

French quantum computing startup Quandela has unveiled Belenos, a 12-qubit photonic quantum computer that it claims delivers 4,000 times more computing power than its predecessor.

The first integrated version is set to be delivered to EuroHPC/GENCI and operated at the CEA’s Très Grand Centre de Calcul (TGCC) before the end of 2025.

Currently, Belenos is accessible via the cloud to over 1,200 researchers across 30 countries in Europe, North America, and Asia.

Instead of relying solely on local deployments, this cloud availability enables researchers to explore hybrid HPC-quantum use cases in fields such as structural mechanics, meteorology, and materials science.

Quandela has ambitious plans to double the qubit count by 2026 with the launch of Canopus. Within three years, the company aims to develop a photonic quantum computer with more than 40 qubits, continuing its focus on systems that avoid cryogenic cooling by using photonics-based methods instead.

‘Our cloud-accessible Belenos system lets partners work on tasks where computing speed and operations per data point are crucial — areas where competitors fall short,’ said co-founder and CEO Niccolo Somaschi.

The platform is designed for practical applications in machine learning and at the AI-quantum interface, which Quandela views as strategically vital sectors for the future.

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

Nvidia recovers as DeepSeek fears fade

Earlier this year, Nvidia shares declined following concerns over DeepSeek and the possibility that tech giants might reduce AI-related spending. Worries over export restrictions added to investor unease.

However, Wedbush Securities’ managing director Matt Bryson believes the DeepSeek issue is now firmly behind the company. According to Bryson, DeepSeek — mostly a China-based phenomenon — unexpectedly boosted demand for AI servers, which ultimately benefited Nvidia instead of hurting it.

Another key development is Oracle’s plan to spend around $40 billion on Nvidia’s GB200 chips to power OpenAI’s new data centre.

Bryson suggested this is part of a broader trend among hyperscalers like Oracle and Crusoe, which recently secured funding to build new facilities. He expects this spending to appear in Nvidia’s earnings as early as Q2 or Q3, instead of being delayed until the next chip generation, the GB300.

Looking ahead, investors remain focused on whether major tech firms will sustain their AI investment. Bryson pointed out that recent earnings reports from companies like Microsoft, Alphabet, and Meta show they remain committed to high capital expenditures.

Instead of retreating, Big Tech appears set to continue driving demand for AI infrastructure, which supports Nvidia’s long-term prospects.

Bryson also noted a significant new factor in AI growth: sovereign deals from countries such as Saudi Arabia and the UAE. He emphasised that the UAE’s expected chip purchases may even surpass Oracle’s.

The new demand, combined with increasing investments in AI-powered edge products — such as those hinted at by OpenAI’s collaboration with Jony Ive — signals that AI spending beyond 2025 will remain strong instead of slowing.

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

Gemini AI powers Aloha 2 robotic demo at Google I/O

At its recent I/O developer conference, Google demonstrated its Gemini AI controlling Aloha 2 robot arms in an interactive exhibit.

The robots were part of the company’s AI Sandbox and offered attendees the chance to interact with the system using voice instructions.

Each setup included a pair of Aloha 2 arms, along with cameras and microphones to process input. The Aloha 2 system, used for teleoperation and data collection, is considered relatively affordable in the robotics field, though a full tabletop setup still costs over $30,000.

The demonstration highlighted Google’s continued push to integrate generative AI into physical systems, combining machine learning with real-world applications.

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

Pakistan aims to become global crypto and AI leader

Pakistan has set aside 2,000 megawatts of electricity in a major push to power Bitcoin mining and AI data centres, marking the start of a wider national digital strategy.

Led by the Pakistan Crypto Council (PCC), a body under the Ministry of Finance, this initiative aims to monetise surplus energy instead of wasting it, while attracting foreign investment, creating jobs, and generating much-needed revenue.

Bilal Bin Saqib, CEO of the PCC, stated that with proper regulation and transparency, Pakistan can transform into a global powerhouse for crypto and AI.

By redirecting underused power capacity, particularly from plants operating below potential, Pakistan seeks to convert a longstanding liability into a high-value asset, earning foreign currency through digital services and even storing Bitcoin in a national wallet.

Global firms have already shown interest, following recent visits from international miners and data centre operators.

Pakistan’s location — bridging Asia, the Middle East, and Europe — coupled with low energy costs and ample land, positions it as a competitive alternative to regional tech hubs like India and Singapore.

The arrival of the Africa-2 subsea cable has further boosted digital connectivity and resilience, strengthening the case for domestic AI infrastructure.

It is just the beginning of a multi-stage rollout. Plans include using renewable energy sources like wind, solar, and hydropower, while tax incentives and strategic partnerships are expected to follow.

With over 40 million crypto users and increasing digital literacy, Pakistan aims to emerge not just as a destination for digital infrastructure but as a sovereign leader in Web3, AI, and blockchain innovation.

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

SynthID Detector aims to boost transparency in AI content

Google has launched SynthID Detector, a verification portal designed to identify whether content was created using its AI models. The tool scans for SynthID, Google’s watermarking technology, which invisibly marks text, images, audio, and video generated by tools such as Gemini, Imagen, Lyria, and Veo.

The Detector highlights which parts of the content likely contain SynthID watermarks. These watermarks are invisible and do not affect the quality of the media. According to Google, over 10 billion pieces of AI-generated content have already been marked using SynthID.

Users can upload files to the SynthID Detector web portal, which then checks for the presence of watermarks. For example, the tool can identify specific segments in an audio file or regions in an image where watermarks are embedded.

Initially rolled out to early testers, the tool will become more widely available in the coming weeks. Google has also open sourced SynthID’s text watermarking technology to allow broader integration by developers.

The company says SynthID is part of a broader effort to address misinformation and improve transparency around AI-generated content. Google emphasized the importance of working with the AI community to support content authenticity as AI tools become more widespread.

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

Claude Opus 4 sets a benchmark in AI coding as Anthropic’s revenue doubles

Anthropic has released Claude Opus 4 and Claude Sonnet 4, its most advanced AI models to date. The launch comes amid rapid industry growth, with the company’s annualised revenue reportedly doubling to $2 billion in the first quarter of 2025.

The Claude 4 models, backed by Amazon and developed by former OpenAI executives, feature improvements in coding, autonomous task execution, and reasoning.

Opus 4 leads in the SWE-bench coding benchmark at 72.5 percent, outperforming OpenAI’s GPT-4.1 and Google’s Gemini 2.5 Pro. Designed for extended task execution, it can maintain focus for up to seven hours, simulating a full workday.

Anthropic says both Opus 4 and Sonnet 4 use hybrid reasoning systems. These allow near-instant responses alongside extended, tool-assisted tasks, including independent web searches, file analysis, and use of multiple tools simultaneously.

Claude models can also build ‘tacit knowledge’ from local file interactions, supporting continuity over time. Sonnet 4, a more efficient alternative to Opus, offers improved instruction following and is already integrated into GitHub’s next Copilot agent.

Both models support expanded developer tools and memory caching through Anthropic’s API, with direct integration into environments like VS Code and JetBrains.

Pricing for Claude Opus 4 is set at $15 per million input tokens and $75 per million output tokens. Sonnet 4 is offered at lower rates of $3 and $15, respectively. Opus 4 is included in Claude’s Pro, Max, Team, and Enterprise tiers, while Sonnet 4 is accessible to free users.

The release also includes Claude Code, a developer assistant capable of reviewing pull requests, resolving CI errors, and proposing code edits. New API features support GitHub integrations, execution tools, and file management.

Anthropic is positioning itself in direct competition with OpenAI, Google, and Meta. While other firms lead in general reasoning and multimodal performance, Anthropic’s strength lies in sustained coding and planning tasks.

However, the company also acknowledged new safety concerns. Claude Opus 4 has triggered Anthropic’s AI Safety Level 3 protocol, following internal findings that it could help users with limited expertise produce hazardous materials.

In response, more than 100 safety controls have been implemented, including real-time monitoring, restricted data egress, and a bug bounty program. Claude Opus 4 and Sonnet 4 are available via Anthropic’s API, Amazon Bedrock, and Google Cloud Vertex AI.

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

Ransomware gang leaks French government emails

A ransomware gang has published what it claims is sensitive data from multiple French organisations on a dark web forum.

The Stormous cartel, active since 2022, posted the dataset as a ‘comprehensive leak’ allegedly involving high-profile French government bodies.

However, researchers from Cybernews examined the information and found the data’s quality questionable, with outdated MD5 password hashes indicating it could be from older breaches.

Despite its age, the dataset could still be dangerous if reused credentials are involved. Threat actors may exploit the leaked emails for phishing campaigns by impersonating government agencies to extract more sensitive details.

Cybernews noted that even weak password hashes can eventually be cracked, especially when stronger security measures weren’t in place at the time of collection.

Among the affected organisations are Agence Française de Développement, the Paris Region’s Regional Health Agency, and the Court of Audit.

The number of exposed email addresses varies, with some institutions having only a handful leaked while others face hundreds. The French cybersecurity agency ANSSI has yet to comment.

Last year, France faced another massive exposure incident affecting 95 million citizen records, adding to concerns about ongoing cyber vulnerabilities.

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

Nvidia ramps up AI push with new Taiwan plans

Nvidia CEO Jensen Huang has urged Taiwan to embrace agentic AI and robotics to tackle its ongoing labour shortage.

Speaking before his departure from Taipei after a week-long visit, Huang said 2025 would be a ‘very exciting’ year for AI, as the technology now possesses the ability to ‘reason’ and carry out step-by-step problem-solving never encountered before.

The new wave of agentic AI, he explained, could assist people with various workplace and everyday tasks.

Huang added that Taiwan, despite being a hub of innovation, faces a lack of manpower. ‘Now with AI and robots, Taiwan can expand its opportunity,’ he said.

He also expressed enthusiasm over the production ramp-up of Blackwell, Nvidia’s latest GPU architecture built for AI workloads, noting that partners across Taiwan are already in full swing.

Huang’s trip included meetings with local partners and a keynote at Computex Taipei, where he unveiled Nvidia’s new Taiwan office and plans for the country’s first large-scale AI supercomputer.

In a TV interview, Huang urged the Taiwanese government to invest more in energy infrastructure to support the growing AI sector. He warned that the energy demands of AI development could exceed 100 megawatts in the near future, stressing that energy availability is the key limitation.

Taiwan’s expanding AI ecosystem — from chip plants to educational institutions — would require substantial support to thrive, he said, pledging to return for Chinese New Year.

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