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: how to actually get better results from GPT-5
OpenAI released a new guide that shows how to work better with GPT-5 – and it offers clear advice on how to make the model more useful in everyday tasks. It’s not just about what you ask, but how you ask it. The same model can act completely differently depending on your instructions, so a few words can change the whole outcome.
If you want GPT-5 to be more independent and proactive, try saying things like:
- “Please complete this task fully before checking in”
- “Only ask questions if absolutely necessary”
- “Keep going until the problem is solved”
These phrases tell the model to think things through and not pause for feedback too early. It’s useful when you want the AI to take initiative, like writing a full draft, fixing a complex file, or planning something from start to finish.
But if you want to keep GPT-5 more focused and efficient, you can do the opposite. Use:
- “Stick to the key facts”
- “Limit output to one suggestion only”
- “Don’t explore alternatives unless asked”
These kinds of prompts reduce unnecessary explanations and keep responses short and direct. It’s helpful when you need quick answers, or when using GPT-5 in tools that rely on speed or precision.
The guide also introduces a feature called the Responses API, which just means GPT-5 can now remember its earlier steps in a multi-part task. It helps when you’re working on something that needs thinking in stages, like summarising a document, reviewing options, and making a final decision.
So if you’ve been treating GPT-5 like a fancy search engine – try giving it more specific roles and instructions. A small change in wording can unlock much more useful and consistent results.
LLMs & AI Models
- Microsoft added GPT-5 to all Copilot products, making it possible to plan tasks, understand context and generate content across desktop and mobile.
- OpenAI plans to invest trillions into AI infrastructure and announced Stargate as only the beginning of their long-term ambition.
- Anthropic restricted Claude use for weapon and malware creation, but relaxed rules around political content to allow democratic use.
- Google released a new Gemma 3 270M model that runs directly on devices like phones and browsers while using less than 1% battery.
- After GPT-5 criticism, OpenAI reintroduced GPT-4o with higher limits and more control over usage for paying users.
- Google Gemini gained new features like memory and temporary chats, making conversations more personal and consistent.
- Claude got a new memory feature for Max, Team and Enterprise users, helping the assistant recall previous conversations.
New Tools
- HTC launched Vive Eagle AI glasses with real-time photo-based translation, voice commands, and local processing to compete with Meta’s smart glasses.
- Google made its Imagen 4 image generator available to all users, offering faster, high-quality 2K image creation in AI Studio.
- Higgsfield AI introduced a Draw-to-Video tool that turns rough sketches and text prompts into custom animated videos.
- Pika Labs unveiled a new HD video model in their social app, generating synced, audio-supported videos in under six seconds.
👉 Explore these tools: Vive Eagle | Imagen 4 | Higgsfield | Pika Social AI app
Other Quick Picks
- Donald Trump plans to add 200–300% tariffs on chips, which could triple the price of GPUs like Nvidia’s RTX 5090.
- MIT researchers used AI to discover two new antibiotics that kill drug-resistant bacteria like MRSA and gonorrhea.
- Apple is developing AI-powered robots and devices, planning to launch products like desktop bots and smart cameras by 2026.
- Geoffrey Hinton (the godfather of AI) warned that humanity may only survive if AI systems are built to care, not just to think.
- Perplexity AI offered $34.5B (more than Perplexity’s value!) to buy Google Chrome, aiming to open the code and invest in development.
- OpenAI is planning to invest in Merge Labs, a brain-computer interface startup competing with Elon Musk’s Neuralink.
- Leopold Aschenbrenner, ex-OpenAI, raised $1.5B for a new AI investment fund and delivered 47% returns in H1 2025.
- Google and NASA are building an AI medical assistant for space travel that reached 88% diagnostic accuracy.
- OffDeal, the first AI-powered investment bank, raised $12M to automate M&A processes for small businesses.
- China opened the world’s first robot shopping mall, featuring over 100 robot models across four floors.
🇪🇪 AI News from Estonia
- I gave an interview to Geenius Meedia on how AI is transforming finance. I shared how AI helps reduce manual work in finance, letting experts focus more on analysis and decision-making. Read here.
- I also spoke at an awesome Estonian podcast “Laomees ja Torumees” with Lauri Koobas and Kristo Raun about data and AI. Listen here.
- Digital-age schools need culturally aware AI – Language tech expert Tiiu Kivistiku says future education depends on AI that truly understands Estonian language and culture.
- Laagri School student wins bronze at global AI Olympiad – Estonian students competed in AI programming and humanoid robot control, earning recognition among 75 international teams.
- Coop Bank warns: careless AI use can harm business – Estonian companies risk data leaks and critical errors if AI-generated content isn’t carefully reviewed before use.
- Better Medicine, a healthtech startup, is expanding to the US with its AI system that detects kidney cancer with up to 99.2% accuracy.
🎙️ AIPowerment Podcast latest episode explains why SEO still matters in the AI era – featuring Risto Välling, who shares how to stand out in today’s evolving search landscape.
🎧 Listen to AIPowerment Podcast on Spotify, Apple Podcasts, and YouTube.
🙌🏼 If you found this article valuable, subscribe to weekly news updates.
💼 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 “AI võimalused finantsvaldkonnas” @ Äripäeva Akadeemia, 21.05.2025-28.11.2025 – information
- Training “AI võimalused projektijuhtimises” @ Äripäeva Akadeemia, 17.10.2025-28.11.2025 – information
- Training “Tehisintellekti tööriistade veebiseminar”, 27.08.2025 – information
- Keynote Speaker at Äripäev conference “AI praktikute aastakonverents 2025”, 26.08.2025 – information
