The US tech giant Microsoft is expanding its AI strategy by integrating Anthropic’s Claude models into Office 365, adding them to apps like Word, Excel and Outlook instead of relying solely on OpenAI.
Internal tests reportedly showed Anthropic’s systems outperforming OpenAI in specific reasoning and data-processing tasks, prompting Microsoft to adopt a hybrid approach while maintaining OpenAI as a frontier partner.
The shift reflects growing strain between Microsoft and OpenAI, with disputes over intellectual property and cloud infrastructure as well as OpenAI’s plans for greater independence.
By diversifying suppliers, Microsoft reduces risks, lowers costs and positions itself to stay competitive while OpenAI prepares for a potential public offering and develops its own data centres.
Anthropic, backed by Amazon and Google, has built its reputation on safety-focused AI, appealing to Microsoft’s enterprise customers wary of regulatory pressures.
Analysts believe the move could accelerate innovation, spark a ‘multi-model era’ of AI integration, and pressure OpenAI to enhance its technology faster.
The decision comes amid Microsoft’s push to broaden its AI ecosystem, including its in-house MAI-1 model and partnerships with firms like DeepSeek.
Regulators are closely monitoring these developments, given Microsoft’s dominant role in AI investment and the potential antitrust implications of its expanding influence.
Would you like to learn more aboutAI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Sam Altman, X enthusiast and Reddit shareholder, has expressed doubts over whether social media content can still be distinguished from bot activity. His remarks followed an influx of praise for OpenAI Codex on Reddit, where users questioned whether such posts were genuine.
Altman noted that humans are increasingly adopting quirks of AI-generated language, blurring the line between authentic and synthetic speech. He also pointed to factors such as social media optimisation for engagement and astroturfing campaigns, which amplify suspicions of fakery.
The comments follow OpenAI’s backlash over the rollout of GPT-5, which saw Reddit communities shift from celebratory to critical. Altman acknowledged flaws in a Reddit AMA, but the fallout left lasting scepticism and lower enthusiasm among AI users.
Underlying this debate is the wider reality that bots dominate much of the online environment. Imperva estimates that more than half of 2024’s internet traffic was non-human, while X’s own Grok chatbot admitted to hundreds of millions of bots on the platform.
Some observers suggest Altman’s comments may foreshadow an OpenAI-backed social media venture. Whether such a project could avoid the same bot-related challenges remains uncertain, with research suggesting that even bot-only networks eventually create echo chambers of their own.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!
OpenAI researchers say large language models continue to hallucinate because current evaluation methods encourage them to guess rather than admit uncertainty.
Hallucinations, defined as confident but false statements, persist despite advances in models such as GPT-5. Low-frequency facts, like specific dates or names, are particularly vulnerable.
The study argues that while pretraining predicts the next word without true or false labels, the real problem lies in accuracy-based testing. Evaluations that reward lucky guesses discourage models from saying ‘I don’t know’.
Researchers suggest penalising confident errors more heavily than uncertainty, and awarding partial credit when AI models acknowledge limits in knowledge. They argue that only by reforming evaluation methods can hallucinations be meaningfully reduced.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
OpenAI is supporting the production of Critterz, an AI-assisted animated film set for a global theatrical release in 2026. The project aims to show that AI can streamline filmmaking, cutting costs and production time.
Partnering with Vertigo Films and Native Foreign, the film is being produced in nine months, far faster than the usual three years for animated features.
The film, budgeted under $30 million, combines OpenAI’s GPT-5 and DALL·E with traditional voice acting and hand-drawn elements. Building on the acclaimed 2023 short, Critterz will debut at the Cannes Film Festival and expand on a story where humans and AI creatures share the same world.
Writers James Lamont and Jon Foster, known for Paddington in Peru, have been brought in to shape the screenplay.
While producers highlight AI’s creative potential, concerns remain about authenticity and job security in the industry. Some fear AI films could feel impersonal, while major studios continue to defend intellectual property.
Warner Bros., Disney, and Universal are in court with Midjourney over alleged copyright violations.
Despite the debate, OpenAI remains committed to its role in pushing generative storytelling. The company is also expanding its infrastructure, forecasting spending of $115 billion by 2029, with $8 billion planned for this year alone.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
The US AI firm OpenAI has introduced a new ChatGPT feature that allows users to branch conversations into separate threads and explore different tones, styles, or directions without altering the original chat.
The update, rolled out on 5 September, is available to anyone logged into ChatGPT through the web version.
The branching tool lets users copy a conversation from a chosen point and continue in a new thread while preserving the earlier exchange.
Marketing teams, for example, could test formal, informal, or humorous versions of advertising content within parallel chats, avoiding the need to overwrite or restart a conversation.
OpenAI described the update as a response to user requests for greater flexibility. Many users had previously noted that a linear dialogue structure limited efficiency by forcing them to compare and copy content repeatedly.
Early reactions online have compared the new tool to Git, which enables software developers to branch and merge code.
The feature has been welcomed by ChatGPT users who are experimenting with brainstorming, project analysis, or layered problem-solving. Analysts suggest it also reduces cognitive load by allowing users to test multiple scenarios more naturally.
Alongside the update, OpenAI is working on other projects, including a new AI-powered jobs platform to connect workers and companies more effectively.
Would you like to learn more aboutAI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Sam Altman, chief executive of OpenAI, has suggested that the so-called ‘dead internet theory’ may hold some truth. The idea, long dismissed as a conspiracy theory, claims much of the online world is now dominated by computer-generated content rather than real people.
Altman noted on X that he had not previously taken the theory seriously but believed there were now many accounts run by large language models.
His remark drew criticism from users who argued that OpenAI itself had helped create the problem by releasing ChatGPT in 2022, which triggered a surge of automated content.
The spread of AI systems has intensified debate over whether online spaces are increasingly filled with artificially generated voices.
Some observers also linked Altman’s comments to his work on World Network, formerly Worldcoin, a project launched in 2019 to verify human identity online through biometric scans. That initiative has been promoted as a potential safeguard against the growing influence of AI-driven systems.
Would you like to learn more aboutAI, tech and digital diplomacy? If so, ask our Diplo chatbot!
OpenAI has reportedly placed a $10bn order with Broadcom to mass-produce custom AI chips, due for shipment in 2026. Sources told the Financial Times that the move would help reduce OpenAI’s dependence on Nvidia, its primary supplier.
Sam Altman recently said OpenAI will use ‘well over 1m GPUs’ by the end of 2025, highlighting the company’s accelerating demand for computing power. In contrast, Elon Musk’s xAI is expected to double its Nvidia Hopper GPUs to around 200,000.
Broadcom confirmed a large custom chip order during its latest earnings call, without naming the buyer. The company’s AI revenue rose 63 percent to $5.2bn, chip sales grew 57 percent to $9.1bn, and shares gained nearly 5 percent.
The new order is expected to be for internal use rather than external customers. Industry observers suggest that OpenAI’s decision signals a strategic shift, allowing the ChatGPT maker to secure supply for its AI expansion while diversifying beyond Nvidia.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
OpenAI has announced plans to launch an AI-powered hiring platform to compete with LinkedIn directly. The service, OpenAI Jobs Platform, is expected to debut by mid-2026.
Applications CEO Fidji Simo said the platform will help businesses and employees find ideal matches using AI, with tailored options for small businesses and local governments. The Texas Association of Business plans to use the platform to connect employers with talent.
The move highlights OpenAI’s efforts to expand beyond ChatGPT into a broader range of applications, including a browser, a social media app, and recruitment. The company faces intense competition from Microsoft-owned LinkedIn, which has been adding AI features of its own.
Alongside the hiring initiative, OpenAI is preparing to pilot its Certifications programme through the OpenAI Academy. The scheme will provide certificates for AI proficiency, with Walmart among the first partners.
OpenAI aims to certify 10 million Americans by 2030 as part of its commitment to advancing AI literacy.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
Despite being promoted as ‘PhD-level’, the model produced a distorted, blob-like map of North America and invented mismatched portraits of US presidents with fictional names.
AI researcher Gary Marcus lowered the threshold by giving GPT-5 a kindergarten-level challenge. The result was a clear fail. He posted: ‘GPT-5 failed a kindergarten-level task. Speechless.’ He criticised the rushed rollout and the hype that may have obscured the model’s visual reasoning weaknesses.
Further tests exposed inconsistencies: when asked to map France and label its 12 most populous cities, GPT-5 returned inaccurate or incomplete results, omitting Paris entirely and naming Orléans despite its lower ranking.
Oddly, when the same queries were posed in text-only form, the model performed better, highlighting the weakness in its image generation and visual logic.
Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!
In 2003-2005, a landmark UN summit – the World Summit on the Information Society (WSIS) – outlined a vision for an inclusive information society, set out recommendations for making this vision a reality, and laid the basis for much of what we call today the global digital governance architecture. Twenty years later, UN member states are looking at progress made in achieving the goals set back then and areas requiring further effort, as well as at whether the WSIS architecture needs updates. This unfolding WSIS+20 review process will end in December 2025 with a high-level meeting of the UN General Assembly. In the lead-up to the meeting, negotiations and consultations are now focused on concrete text for what will become a WSIS+20 outcome document. This concrete text – called the zero draft – was released last week.
Digital divides and inclusion take centre stage in the zero draft. While connectivity has expanded – 95% of the global population is now within reach of broadband, and internet use has grown from 15% in 2005 to 67% in 2025 – significant gaps remain. Disparities persist across countries, urban and rural areas, genders, persons with disabilities, older populations, and minority language speakers. The draft calls for affordable entry-level broadband, local multilingual content, digital literacy, and mechanisms to connect the unconnected, ensuring equitable access.
The digital economy continues to transform trade, finance, and industry, creating opportunities for small and women-led businesses but also risks deepening inequalities through concentrated technological power and automation. Against this backdrop, the draft outlines a commitment to supporting the development of digital financial services, and a call for stakeholders to foster ‘open, fair, inclusive and non-discriminatory digital environments.
Environmental sustainability is a key consideration, as ICTs facilitate monitoring of climate change and resource management, yet their growth contributes to energy demand, emissions, and electronic waste. Standing out in the draft is a call for the development of global reporting standards on environmental impacts, and of global standards for sustainable product design, and circular economy practices to align digital innovation with environmental goals.
The Zero Draft reaffirms human rights, confidence and security, and multistakeholder internet governance as central pillars of the digital ecosystem. Human rights are positioned as the foundation of digital cooperation, with commitments to protect freedom of expression, privacy, access to information, and the rights of women, children, and other vulnerable groups. Strengthening confidence and security in the use of technology is seen as essential for innovation and sustainable development, with emphasis on protecting users from threats such as online abuse and violence, hate speech, and misinformation, while ensuring safeguards for privacy and freedom of expression.
The draft outlines a series of key (desirable) attributes for the internet – open, free, global, interoperable, reliable, secure, stable – and highlights the need for more inclusive internet governance discussions, across stakeholder groups (governments, the private sector, civil society, academia, and technical communities) and across developed and developing countries alike.
To advance capacity building in relation to AI, the draft proposes a UN AI research programme and AI capacity building fellowship, both with a focus on developing countries. In parallel, the draft welcomes ongoing initiatives such as the Independent International Scientific Panel on AI and the Global Dialogue on AI Governance.
Recognising the critical importance of global cooperation in internet governance, the draft designates the Internet Governance Forum (IGF) as a permanent UN body and calls for enhanced secretariat support, enhanced working methods, and reporting on outcomes to UN entities and processes (which are then called to duly take these outputs into account in their work). The long-discussed issue of IGF financial sustainability is addressed in the form of a request for the UN Secretary-General to make proposals on future funding.
Finally, the draft looks at the interplay between WSIS, the Global Digital Compact and the 2030 Agenda for Sustainable Development, and outlines several mechanisms for better connecting them and avoiding duplication and overlaps. These include a joint WSIS-GDC implementation roadmap, the inclusion of GDC review and follow-up into existing annual WSIS mechanisms (at the level of the Commission on Science and Technology for Development and the Economic and Social Council), and reviews in GDC-WSIS alignments at the GA level Speaking of overall reviews, the draft also envisions a combined review of Agenda 2030 and of outcomes of the WSIS-GDC joint implementation roadmap in 2030, as well as a WSIS+30 review in 2035.
Looking ahead
The Zero Draft sets the stage for intense negotiations ahead of the December 2025 High-Level Meeting. Member states and other stakeholders are invited to submit comments until 26 September. It then remains to be seen what a second version of the outcome document will look like, and which elements are kept, revised, or removed.👉 Follow the process with us on our dedicated web page, where we will track key developments, highlight emerging debates, and provide expert analysis as the negotiations unfold.
DW Team
Highlights from the week of 29 August – 5 September 2025
The EU General Court upheld the EU–US Data Privacy Framework, rejecting claims it lacks adequate safeguards and independence in oversight of US data practices involving personal data from the EU.
The document flagship outlines projects in AI, quantum, cloud, and space, promotes a Franco-German digital ecosystem for public services, and sets the stage for the 2025 European Digital Sovereignty Summit.
A joint cybersecurity advisory details how Salt Typhoon exploited unpatched network-edge devices to infiltrate telecommunications, military and government systems across 13 countries.
The SCO Tianjin Declaration emphasised cyber sovereignty, inclusive AI development, global cybersecurity norms, and stronger cooperation in the digital economy.
As classrooms and negotiation tables fill again, a pressing question lingers: where is AI headed? This summer marked a turning point, as the ‘bigger is better’ narrative faltered. This blog captures ten key lessons from a season of AI disillusionment.
Why apprenticeship and storytelling are the future of learning in the AI Era AI is forcing us to ask a deeper question: what is the real purpose of learning?
The webinar will bring together African experts from technology, development, diplomacy and policy domains to discuss which digital issues must be urgently prioritised to keep Africa on course in a rapidly changing world.
In this one-hour session, several experts – Asoke Mukerji, Isaac Morales Tenorio, and Fan Yang will debate the future of global cyber negotiations — tackling obstacles, testing new ideas, and asking whether the UN dialogue can move from compromise to real progress.
The event will discuss the progress made by the Multi-Stakeholder Working Group on Data Governance and the expectations for the next meeting, which will take place on 15-16 September.
