He suggests that education could be a way to avoid geopolitical competition in AI development.

As states assert control over data flows, digital platforms, and taxation, the model of an open and interoperable internet risks being eclipsed by competing, state-centric visions of digital order.

As more countries erect fiscal and regulatory barriers in the name of digital sovereignty, the once-unifying dream of a borderless, interoperable internet is giving way to a fragmented digital terrain defined by jurisdictional silos and incompatible standards.

confrontation over compromise, national advantage over global consensus, and short-term leverage over long-term stability.

In February 2025, Trump’s administration escalated further, threatening to invoke Section 891 of the U.S. tax code—an obscure statute that would double corporate taxes on firms from countries taxing U.S. digital companies.

Retaliatory tariffs under Section 301 of the U.S. Trade Act were imposed on countries implementing DSTs, including France, the UK, Italy, Spain, Turkey, and Austria.

multinational digital corporations contribute fairly to local economies, especially in jurisdictions where they generate substantial revenue without a physical presence.

Most Popular AI Tools

Promoting Registered Apprenticeships.  

can integrate the fundamentals of AI into all subject areas

reducing time-intensive administrative tasks;

15 U.S.C. 9401(3).

invest in our educators

demystifies this powerful technology

using stolen U.S. technology on the back of U.S. semiconductorchips that are prohibited from sale to China without an export license

a model that covertly censors and manipulatesinformation pursuant to Chinese law

AI Foundation

European AI Champions Initiative

Industries will be able to collaborate and federate their data

CERN laboratory in Geneva

with AI Gigafactories. Very large data and computing infrastructure, to train very large models.

12 AI factories

European AI embraces the power of open source

European AI is cooperative. It brings talents together from different countries, sectors and backgrounds.

Because the AI race is far from over. Truth is, we are only at the beginning. The frontier is constantly moving. And global leadership is still up for grabs.

AI can help us boost our competitiveness, protect our security, shore up public health, and make access to knowledge and information more democratic.

And this means embracing a way of life where AI is everywhere.

We want Europe to be one of the leading AI continents

accelerat

OpenAI, Deepmind, Anthropic, Meta, and Deepseek don’t need to use a standardized neural architecture so that their AIs can talk to each other.

Overall, the Internet and AI are different technologies on different layers of the tech stack.

the nodes of a computer network

“The” Internet – singular – is a global communication network.

The Internet is a communication protocol for sending data packets between computers in a standardized way. AI is an information technology that processes inputs and creates outputs.

How can companies demonstrate a multipolar mindset on people? Here are a few practices CEOs should consider:

Geopolitics is not only personal—it’s emotional.

a culture of geopolitical resilience

using advanced technologies to anticipate and prepare for future geopolitical disruptions.

One private equity fund, for instance, identified portfolio companies in regions where profits were likely to shrink because of regional conflicts and redirected them to more stable geographies. Additionally, fund executives identified products that could be considered “dual use”—that is, products that could have both commercial and military applications—and would be especially vulnerable to cross-border trade scrutiny. They relocated the manufacturing of these products to areas that would be less exposed to dual-use regulations if those regulations were to increase in scope.

It will be critical for business leaders to monitor where this funding lands.

They are finding opportunities in three areas in particular—accelerating growth, optimizing business operations, and developing capabilities and strategies to address global disruption.

proactive approach to geopolitics

Exhibit 1

These include tariffs, the provision of subsidies in support of national industrial policies, and governments’ bias toward investing in geopolitical allies across supply chains, talent, technology and data, capital deployment, and other business domains.

Business leaders should be asking themselves questions such as, will our competitors’ products be more or less expensive than ours because of new tariffs and taxes? When and how can we align our business with trade flows into new corridors? What new economic and security alliances could also create opportunities for us to grow or to change our cost structure? What industrial policy incentives might present significant growth potential for us? How is our risk-adjusted cost of capital changing across geographies and how might we optimize our capital deployment?

Serving as our “technologies of memory,” our grandmothers and other cultural storytellers have been responsible for preserving and passing down stories of our history and culture long before the advent of the printing press or the internet.

Many of the raw materials required for AI infrastructure, including cobalt, copper, and lithium, are sourced from areas in the Global South, including countries facing ongoing conflicts or near lands where Indigenous communities are located, adversely impacting the environment for those communities.

A key piece of the pie in creating an “age-aware internet” is age assurance. Iain Corby, director of the Age Verification Providers Association (AVPA), says 2025 is set to be a busy year in regulations, with Ofcom expected to publish this week their guidance on “highly effective age assurance.”

For your second concern, we compare HtmlRAG with chunking parctice in Langchain in our paper. We are talking about an RAG system whose knowledge source is basically in HTML format. Those fancy chunking strategies are based on the HTML-to-plain-text conversion. If information loss is mainly brought by the conversion other than the chunking strategy, different chunking strategies may have little difference.

Great suggestion👍! Actually, for each block, there are many additional features waiting to be explored, such as tag attributes, url links, and context path you have mentioned. We can probably optimize the block represention strategy in future works.

the future of social media and the relationship between American business and government.

Yes. Money is a story, a fiction—it has no objective value.

We build trust by inventing stories that many people believe.

Managers using agentic AI systems will need to pay careful attention to team selection, ensuring that they have the right combination of agentic roles carrying out the right tasks, in an efficient way.

contextual information

In fact, goal-setting becomes even more important for agentic AI, as the systems initially lack the contextual information — such as organizational and market context, company values, and so forth — that is often tacitly understood by human workers.

Ema, an AI startup based in California, offers agentic AI chatbots that can dynamically trawl thousands of different databases and apps to resolve customer queries and complaints, learning from each customer interaction and identifying recommended actions for human agents. Ema also audits its content for accuracy and compliance purposes, while also making recommendations to improve the customer knowledge base.

SciAgents — a multi-agent model developed by researchers at MIT — includes not only robot scientists to develop research plans, but a Critic Agent to review these and suggest improvements.

Because agentic models are explicitly designed to carry out very granular tasks, they enable much greater specialization of roles compared with previous broad-brush automation systems.

While agentic AI systems harness the creative abilities of generative AI models such as ChatGPT, they differ in several ways. First, they are focused on making decisions rather than on creating content. Second, they do not rely on human prompts, but rather are set to optimize particular goals or objectives, such as maximizing sales, customer satisfaction scores, or efficiency in supply-chain processes. And third, unlike generative AI, they can also carry out complex sequences of activities, independently searching databases or triggering workflows to complete activities.

In April the Trade and Technology Council, a forum for co-operation between America and Europe on economic issues, released a statement saying that democracies must remain the vanguard of emerging technologies, including by setting the standards that underpin them.

It’s 55% yes and 44% no.

The meaning of the word “squirrel” comes from its PLACE within this web of distinctions.

FORGET UNITY with a whole altogether…it is POSSIBLE to be INTERDEPENDENT with EVERYTHING around you for what you are…but NOT interdependent with EVERYTHING there possibly is.

See it’s NOT that the self is a TOTAL illusion to him. The DURABILITY of the self is an illusion…the ESSENCE at the FOUNDATION of the self is an illusion. 

ou realize the typical boundaries between YOU and the eco-systems of ideas or things that you depend on…are really just a matter of convention… or a matter of the ways that our languages typically manage to express things. 

The WAY we talk about things…USUALLY tries to say something about what things ARE…by COMPARING them to the things that they’re not— our languages STRUCTURE our reality dualistically with subjects always relating to objects. 

there is some sort of static essence or form to what a Nihilistic experience will be– and TWO…it removes the expectation that we should be able to perfectly describe ALL of this experience using purely language.

no two encounters with Nihilism are EVER going to be the same… and what you experience and CALL Nihilism… will without question be unique to your individual experiment. 

say after the fall of the Roman Empire, or the French Revolution and the Reign of Terror…

And what he’s ULTIMATELY doing here in the context of post-ww2 Japan, a time and place where the culture is facing EXTREME Nihilism after the dropping of the atomic bomb, the failure of the imperialism of early 20th century Japan…when they’re doing some SOUL searching in Japan about where to go to NEXT…Nishitani is a thinker that’s looking to OTHER cultures that have gone through something similar in the past…the HOPE being to learn from them by radically engaging with their work. 

modern society often SHIELDS people from it. 

including top oil and gas producers like Saudi Arabia and the US, acknowledged the need to phase down fossil fuels.

“With no real money on the table, and vague and unaccountable promises of funds to be mobilised, they are trying to shirk their climate finance obligations,” he added, explaining that “poor countries needed to see clear, grant-based, climate finance” which “was sorely lacking”.

of $1.3 trillion annually by 2035 was adopted

If you miss anything, don’t worry: you will be able to read through the transcript of this meeting after it ends. You will find the transcript posted in the classroom, in the course outline, under the Orientation module.

Textus annotations are used for asking questions, posting comments, expressing your experiences/ideas/observations/, and sharing links to resources directly related to parts of the lecture texts. They are our main tool for discussion of the course materials with your classmates and the course team.

When you open the module reading, all annotations will appear highlighted in yellow. Click on the highlight, and the textus menu will open on the right-hand side of the classroom, where you’ll both see the annotation and the option to respond to it.

Via the new annotations alert icon in the classroom (in the upper right-hand side corner, next to your profile icon) that shows annotations made since the last time you logged in. Click on it to open a small drop-down menu with the list of new annotations, and then click on each of them to view it in the textus menu.

This is what I planned to cover today! I’m glad to see you've caught on very quickly how to use the textus annotation tool. Now you have the ability to go ahead with the tasks for the week!

The shift in Meta’s and Anthropic’s policies toward military uses of AI is momentous. Lately AI has become a battlefield staple in the form of weaponized drones, and AI companies must take care that their new policies are consistent with upholding human rights. Military uses for AI include not only weapons development and targeting but also potentially life-saving search and rescue, logistics, intelligence, and communications. Moreover, defense contracts represent major opportunities for AI companies that can fund widely beneficial research and applications.

Technology transfer: Developing and least developed countries (LDCs) may require technology transfer to build the capacity of local companies.

If source code is unlawfully copied, obtained, or used dishonestly for commercial gain, WTO members may pursue legal action against the offender in the courts of another member.

The source code of computer software is protected in the same way as a ‘literary work’, meaning it is under copyright protection from the moment the first line of code is created.

Fostering trust among actors is also essential

strongly encourage American tech companies to build their largest and most advanced facilities in the United States.

the weights of frontier models (the parameters that encode the core intelligence of an AI system).

Zelenskyy said in his evening address to Ukrainians.

each AI system ethical performance can be assessed from the very outset.

with hands-on

There is clearly a tendency for ideological fundamentalists to present the world as a struggle between good and evil, in which mutual understanding and pragmatism are demonized as a betrayal of sacred values. 

“We have built a foreign policy of hectoring and moralizing and lecturing countries that don’t want anything to do with it.

If the US can accept a more modest role in the international system as one among many great powers, the president elect could embrace a more moderate economic nationalism that would have a greater prospect of succeeding.

Globalization gave birth to a political class loyal to international capital without national loyalties, and accountability to the public disappeared. Globalization often contradicts democracy, and there is a growing division between illiberal democracy versus undemocratic liberalism.

Developers are continuously iterating on their prompts so that their applications achieve more desirable outputs from the LLMs they are calling.

Vector databases treat embeddings as independent data, divorced from the source data from which embeddings are created, rather than what they truly are: derived data. By treating embeddings as independent data, we’ve created unnecessary complexity for ourselves.

hatGPT has empowered current AI algorithms to achieveunprecedented levels of strength and effectiveness, reshaping the way humans employ or develop AI algorithms.Its emergence has captured the attention of the research community.

ActiveX

line le

UNGA79 and the ‘Pact for the Future’

The ink on the Pact for the Future was barely dry

a global handshake

for the unpredictable rhythms of modernity.

The ‘Pact for the Future,’ adopted at the Summit of the Future on 22 September 2024, sets out an ambitious agenda to address climate change, digital transformation, and peace while fostering agile global governance. 

Gold has soared to a record high of $2,629 per ounce following the US Federal Reserve’s recent interest rate cut.

AI models, including ChatGPT and Cohere, once depended on low-cost workers for basic fact-checking. Today, these models require human trainers with specialised knowledge in medicine, finance, and quantum physics. 

While South Korean memory giants Samsung Electronics and SK hynix experienced a significant sales increase in China during the first half of this year, the report by the Korea Eximbank Overseas Economic Research Institute indicates that South Korea’s reliance on China for critical semiconductor raw materials is also growing.

a lighthouse – navigating

Moscow’s efforts have focused on supporting Donald Trump and undermining Kamala Harris.

to leap from the past into an uncertain, but ambitious, tomorrow.

Fighterbomber

was lost due to the absence of recordings, transcripts, or even note taking.

Gae down by Faile, and taste the ale,

Everyone makes hundreds of decisions a day, but most of the time they aren’t conscious of the decisions they are making.

The first is the geostrategic importance of technology – in particular data consuming and content producing technologies (which explains why the US has been so keen to limit the power of China’s TikTok).

I suspect for instance there will be a lot more civil and corporate legal cases to establish the ownership of data-sets and the access to them by AI engines, as is the case between the New York Times and OpenAI.

technology weighs on many public policy debates now – such as the access to mobile phones in schools, the demands that data centres place on electricity grids and the TikTok-ification of politics.

Productive stupidity means being ignorant by choice. Focusing on important questions puts us in the awkward position of being ignorant. One of the beautiful things about science is that it allows us to bumble along, getting it wrong time after time, and feel perfectly fine as long as we learn something each time. No doubt, this can be difficult for students who are accustomed to getting the answers right. No doubt, reasonable levels of confidence and emotional resilience help, but I think scientific education might do more to ease what is a very big transition: from learning what other people once discovered to making your own discoveries. The more comfortable we become with being stupid, the deeper we will wade into the unknown and the more likely we are to make big discoveries.

Like other comments, I was also initially surprised. But I think the gains are both real and easy to understand where the improvements are coming from.

Chain-of-Thought and Self-Reflection are also very simple, and this was built with a SYNTHETIC dataset. Can you imagine if we used people to carefully curate a much better dataset? If we experimented with other forms of CoT, self-reflection, multi-agent reasoning, etc? There are so many things to try, and you don't have to be at OpenAI to try them!

Imagine a text chunk including a sentence like “Donald Trump says that there should be a wall on the southern border of the US.” The meaning of this sentence could not be understood unless one knows the broader context—such as if Donald Trump was incumbent when he made such a statement. If he was, then the sentence probably conveyed the official stance of the US at that time; otherwise, it would be Mr. Trump’s personal stance. The Manager argued that an RM model—such as the one used by the lab at that time—only conducted semantic searches and could not capture context beyond the text chunk; it might mistake official country stances, retrieve the wrong paragraphs for the chatbot, and lead the user to believe in wrongful answers.

The way that the AI reporting system should transform information into knowledge now includes comparing current knowledge to historical knowledge,

It turned out that, to her, the most valuable information would not be what was said this year but what was said this year that was different from all previous years.

The AI lab brought in a linguist, who broke down conversations into questions and answers; the linguist taught the AI lab various question types one could pose in a conversation and ways to detect which type they were (rhetorical, open, etc.). The next step was to understand the answers; the AI lab consulted a debater who framed responses in terms of arguments, which were then understood as key points with corresponding supporting facts. Taking in these lessons, the AI lab instructed the reporting system to take each speaker’s paragraphs from a transcript, extract key points and supporting evidence, and present a session summary in this format.

The course team prepares various courses that the Foundation delivers to diplomats or higher-education students. The reporting team takes care of the daily monitoring and updating of digital news and global policy trends on the Foundation’s website; it is also the team that used to provide live coverage of major international events, such as the UN General Assembly (UNGA) and Internet Governance Forum (IGF). The creative lab designs social media campaigns and visuals for all published material. The tech team manages the technical infrastructure, from websites to internal tools and applications, that allows the Foundation to function.

AI models are not merely machines that solve our problems; they are our imagination of the world and its problems; they are our respective worldviews, personalities, ambitions, and desires reformulated, reconfigured, and translated into codes.

In the Director’s envisioned knowledge management system, the technology product must not be a mere standalone tool but a part of a workflow that processes information

The AI reporting system takes an audio-visual recording of a session, transcribes all speakers’ speeches via a transcription model (TM), and then generates summaries of various formats and knowledge graphs based on the transcripts via several summarization models (SMs) and a knowledge-graph generating model (KGGM).

We recognize the importance of  continuing to pursue governance reforms

Trade provides another example of the failure of multilateral negotiations (Action 5). The WTO (which has replaced UN Trade and Development, or UNCTAD, in this function) has rules that are being willfully ignored. The US has knee-capped the dispute mechanism by refusing to allow the appointment of judges; the world’s largest exporter, China, continues to claim developing-country status and hides information on its trade policies. Global trade negotiations began in 2001 in Doha and died in 2015. Determined multilateralists are calling for WTO reform but there is no sign that the changes will happen. There are echoes here of the ill-fated efforts to change the Security Council.

But the COPs have been singularly ineffective for several reasons. One is the UN secretariat’s timidity. It has been far too indulgent to member states in allowing voluntary national targets to reduce carbon emissions. It has permitted petro-states to host COP meetings, like the United Arab Emirates last year and Azerbaijan this year, and it has failed to counter the advocacy by oil and gas interests. Progress is measured by slight adjustments of language in its outcome statements while downplaying actual outcomes in CO2 emissions.

the First UN of member states — and particularly the veto wielding P3 of Russia, China and the United States — and the Second UN of secretariats.

Respect for human rights, self-determination, territorial integrity, freedom of navigation, democratic governance, free movement of goods, economic openness etc. are important values that may be invoked by a court to assist it in the interpretation of legal rules but they are not rules of law as commonly understood.

the narrow positivist perception of international law to include soft law, including the standards and recommendations of international standard-setting organizationsFootnote 17 and conferences and rules made by non-state actors.

Linguistic research had identified 80,000 “noun synonym sets”, or synsets: groups of synonyms that described the same sort of thing.

calls the tech giant’s models “open weights”, rather than open source.

On one side are the open-source purists in favour of decentralised “little tech”. On the other are closed-source realists who argue that greater centralisation and control are better for safety and national security.

Starlink, a satellite-internet firm, could be a stopgap, but regulators have blocked it in at least seven countries including South Africa.

Mobile-phone services directly contribute $170bn to Africa’s gdp.

We’re testing SearchGPT, a temporary prototype of new AI search features that give you fast and timely answers with clear and relevant sources.

Impressive AI video generator KLING now available as web version

a paper introducing OpenDevin, a framework for developing AI agents that interact with environments similar to human programmers.

It’s OpenAI, Google, and Anthropic vs. Meta, Llama, and xAI. One thing is for certain: open-source big AI is going to be competitive.

A 2009 agreement insisted on by the European Commission meant that Microsoft could not make security changes that would have blocked the update from cybersecurity firm Crowdstrike that caused an estimated 8.5 million computers to fail, the Big Tech giant said in comments to the Wall Street Journal newspaper.

Modi sent his foreign minister, Subrahmanyam Jaishankar, to the Shanghai Cooperation Organisation (SCO) at its annual meeting in the Kazakhstan capital, Astana, which was also attended by Putin and the Chinese president, Xi Jinping.

“expert systems” which used symbolic logic to capture and apply the best of human know-how.

Amos Hochstein, Joe Biden’s envoy for energy security, hopes that, by redirecting exports to the Atlantic, supply chains will shift decisively westward.

the work

Superintelligence (also known as Artificial Super Intelligence, or ASI) is defined much more clearly as an AI having intelligence ‘far surpassing that of the brightest and most gifted human minds’.

likely

undefined marketing term

most people

simulating reasoning, avoiding hallucinations, and enhancing the capacity to use external tools.

Join our community of global citizens empowered by The New Humanitarian’s independent journalism from the heart of crises, conflicts, and disasters.

tested this idea by “spiking” (ie, turning up) a feature associated with the Golden Gate Bridge.

“mechanistic interpretability”.

“lots of different brains looking at the same source at the same time”.