The most important news pieces from the AI world are here for you, whether you’re an AI enthusiast or just wish to keep an eye on what’s going on.
Focus Topic: A global snapshot of AI use in 2025: Microsoft’s new report shows who's ahead and who’s falling behind
Microsoft’s AI Economy Institute just released a new report that provides the clearest global picture yet of how fast generative AI is spreading – and where it’s not. The data shows that global usage hit 16.3% of the population in the second half of 2025 – up from 15.1% earlier that year. That means about one in six people now use generative AI to work, learn or create. But the report also highlights a growing concern – the digital divide is widening, not closing.
While usage in the Global North jumped from 22.9% to 24.7%, growth in the Global South lagged behind at just 14.1% – expanding the North-South gap from 9.8 to 10.6 percentage points. The fastest adopters are nations that invested early in digital infrastructure and policy, such as the UAE, Singapore, Norway and Ireland. The UAE now leads globally, with 64% of its working-age population using AI tools, ahead of Singapore (60.9%) and Norway (46.4%).
A standout story is South Korea, which leapt seven positions to 18th place globally, thanks to government-led AI policies, education programs, and improved Korean-language support in large models. GPT-5 and GPT-4o dramatically boosted performance in Korean, and viral trends like Ghibli-style AI images helped trigger wider use. In just a few months, South Korea saw an 80% increase in AI usage since October 2024 – now the second-largest ChatGPT market worldwide.
Meanwhile, a new player is emerging in underserved regions. DeepSeek, a Chinese open-source AI model, is gaining traction across Africa, Russia, and Iran – thanks to its zero-cost access and partnerships with Huawei. In some African markets, DeepSeek usage is 2-4× higher than in Europe or North America.
The report makes one thing clear – the next billion AI users won’t come from Silicon Valley, but from regions where cost, access, and language support drive adoption. The challenge ahead is not just building smarter AI – it’s ensuring that access spreads faster than inequality.
LLMs & AI Models
- OpenAI launched ChatGPT Translate, a standalone AI translator that lets users fine-tune results for tone, audience, or context, but lacks full feature parity with Google Translate. You can find it here: https://chatgpt.com/translate/.
- OpenAI began testing ads inside ChatGPT for free and Go users in the US, promising they won’t affect conversations and are clearly marked.
- Custom GPTs now get real-time access to company tools like calendars and docs without code or manual uploads, replacing the older Actions system. Currently only in the US with limited business accounts.
- Anthropic launched Claude Labs, a new unit focused on experimental features and advanced capabilities, led by ex-Instagram and Meta AI talent.
- Claude Cowork, a new macOS tool, brings agentic capabilities to daily tasks to Max users – managing files, reports, and tasks across Asana and Notion for non-developers.
- Claude for Healthcare and Life Sciences got upgrades – now supporting data sources like ICD-10 and ClinicalTrials.gov for claims, protocols, and health data.
- Anthropic cut xAI’s access to Claude models after finding misuse via Cursor, a violation of service terms confirmed by xAI’s leadership.
- Google Gemini added Personal Intelligence, letting AI use your Gmail, Photos, and YouTube for context-aware answers, though the feature is off by default.
- Google and Apple announced a multi-year deal to bring Gemini into Siri, while keeping privacy features like on-device compute in place.
- Thinking Machines lost three co-founders to OpenAI, amid internal leaks and leadership changes; PyTorch creator now named as new CTO.
New Tools
- Replit now lets users build, test, and publish mobile apps directly to the App Store – all from one platform using QR code previews.
- Zhipu AI released GLM-Image, a new open-source image model trained fully on Huawei chips, marking a step toward China’s chip independence.
- Google released Veo 3.1, a tool for creating 4K vertical AI videos, now integrated with Gemini, YouTube Shorts, and Google Vids.
- Slack’s Slackbot became a personal AI agent, using messages, channels, and files to respond with better context and support workplace workflows.
- Manus launched Meeting Minutes, a tool that records physical meetings and creates task-based summaries to support structured follow-ups.
- ElevenLabs introduced Scribe v2, a new speech-to-text model claiming industry-low error rates and SOTA accuracy across 90+ languages.
Other Quick Picks
- Cursor agents built a full web browser using hundreds of autonomous GPT-5.2 agents – over 3 million lines of code in under a week.
- OpenAI invested in Merge Labs, a brain-computer interface startup using non-invasive tech to read brain data with AI, a competitor to Elon’s Neuralink.
- Allie K. Miller shared her 2026 AI stack – no universal model exists, best results come from mixing tools like Claude, Gemini, and Nano Banana.
- Matthew McConaughey trademarked his voice and likeness to fight deepfakes – now legally protected including catchphrases and clips.
- Meta announced a 600B USD AI infrastructure plan, starting in the US and fueled by nuclear energy, with 20-year energy contracts.
- Microsoft launched a community-first AI data center initiative, promising no energy price hikes and full property tax payments.
- McKinsey now uses 25,000 AI agents alongside human staff – 40% of company roles are already AI-assisted or AI-driven.
- Google introduced the Universal Commerce Protocol, an open AI framework to manage full shopping journeys inside AI chats.
🇪🇪 AI News from Estonia
- Luukas Ilves built Slipup Finder (Apsakaleidja), an AI tool that scans laws for logical and technical errors like inconsistent definitions and broken references. Test it out here.
- Gustav Adolf Gümnaasium students said AI helps them save time and understand material better – especially for tests and summaries. Watch interview here.
- Flashka raised €1M pre-seed led by Outlast Fund to scale its AI learning engine and expand across Europe.
- Hansa Grupp merged CEO and CTO roles to align on a new data-first, AI-ready strategy – 2026 is “data year”, 2027 will be AI year.
- Doris Põld shared an interesting 2026 forecast on Estonian IT sector with references to Gartner and other research. Read more here.
- Telia reports Estonia faced 2.9M cyberattacks per day, most launched by AI-powered tools; human response still key to defense. Read here.
🎙️ AIPowerment Podcast dropped a new episode featuring Gerlyn and Sandra, discussing 2026 AI trends and predictions – Agent-to-Agent interactions, vibecoding, physical AI hardware, single model trap, and lots more.
🎧 Listen to AIPowerment Podcast on Spotify, Apple Podcasts, and YouTube.
Want to stay in the loop?
Straight to your inbox: practical AI updates, finance use cases, tools to try, and upcoming trainings.
I cut the noise and send only what’s actually useful.
💼 Exploring how to make AI work in finance? Let’s talk practical use cases – connect with Gerlyn Tiigemäe for expert guidance.
📚 If you would like to participate in one of my trainings or listen to speaking engagements, here are the upcoming ones:
- Training “Tehisintellekti (AI) kasutamine finantsanalüüsiks” @Eesti Koolitus- ja Konverentsikeskus, 21.01.2025 – information
- Training “AI võimalused finantsvaldkonnas” @ Äripäeva Akadeemia, next trainings in 2026 – information
- Training “AI võimalused projektijuhtimises” @ Äripäeva Akadeemia, next trainings in 2026 – information
- Training “AI assistendi roll ja võimalused”, 03.02.2026 – information
