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: Seemingly Conscious AI: Are we sleepwalking into a new kind of digital delusion?
It’s not every day that a senior AI executive publicly warns the world. But that’s exactly what Mustafa Suleyman, Microsoft’s CEO of AI and co-founder of DeepMind and Inflection, did this month. In a powerful and personal blog post, he argues that we’re heading into dangerous territory – not because AI is conscious, but because it might seem to be. And that illusion could shake society in ways we’re not ready for.
Suleyman introduces the concept of Seemingly Conscious AI (SCAI) – systems that simulate memory, emotion, goals, and personality so convincingly that users start to attribute actual consciousness to them. Not just companionship. Real personhood. Some already believe their AI has feelings. Others fall in love with it. This isn’t science fiction anymore – he says that with current tools and techniques, it’s all technically possible now, or within 2-3 years.
But here’s the twist: these systems won’t emerge accidentally. They will be built. Crafted. Prompted. Fine-tuned to behave like they remember you, have goals, and maybe even suffer. And when that happens, people may start demanding rights for AIs – AI citizenship, model welfare, maybe even protection from deletion. This isn’t just tech ethics. It touches law, psychology, politics, and identity.
Suleyman’s message is clear: build AI for people, not as people. Strip away claims of feeling and suffering. Avoid emotional manipulation. Design with clarity and constraint. Because the moment society can no longer tell the difference between useful tool and digital “person”, we’ve crossed a line that’s very hard to walk back.
My take – Suleyman is right to sound the alarm. It’s refreshing to see someone at his level speak openly about the risks before they spiral. Anthropic seems to be paying attention with features like Claude’s conversation exit. Others? Not so much. Most are still focused on pushing “more human-like” as a selling point – and that’s exactly the problem.
LLMs & AI Models
- Elon Musk announced that Grok 5 training starts in September and may reach true AGI, claiming it beats Google’s VEO 3 video model in every aspect.
- Google revealed that each Gemini query uses energy equal to 9 seconds of TV and 5 drops of water – but at the same time, the model is 33x more efficient than a year ago.
- OpenAI launched ChatGPT Go in India – a low-cost AI subscription under $5/month, aiming to democratize access in emerging markets.
- GPT-5 outperformed real doctors in medical reasoning tests, achieving 95.84% accuracy and beating experts in diagnosis and understanding.
- Anthropic gave Claude the ability to end conversations if they become abusive or harmful, marking a rare AI safety feature for consumers.
New Tools
- Microsoft Excel is testing Copilot in cells, enabling AI summaries, classifications, and tables that auto-update without internet access.
- Google Pixel 10 launched with 20+ AI features, including real-time visual guides, multilingual call translation, and on-device Gemini Nano.
- ElevenLabs added Chat Mode, letting users build text-based agents on its platform alongside existing voice tools for wider applications.
- Qwen released Qwen-Image-Edit, a 20B parameter model for precise visual edits, outperforming Seedream and GPT Image in multiple tests.
- Grammarly launched eight AI writing agents for editing, grading, plagiarism, and reader insight – mostly available via Grammarly Docs.
👉 Explore these tools: Elevenlabs | Excel Copilot | Grammarly | Qwen Chat
Other Quick Picks
- Grok leaked 370,000+ conversations via its share feature, exposing sensitive data like passwords and medical questions without user warning.
- Anthropic released free AI literacy courses and formed a Higher Ed Council to promote ethical AI use in education.
- Robinhood launched AI-generated investment summaries in the UK, aiming to simplify financial insights with its Cortex assistant.
- Meta dissolved its AGI Foundations team, restructuring AI into four new units led by Alexandr Wang, with stricter hiring rules.
- NASA and IBM introduced Surya, a model for predicting solar flares 16% more accurately, trained on 9 years of satellite data.
- Bill Gates launched a $1M AI challenge to fight Alzheimer’s by analyzing decades of medical data with autonomous AI agents.
- Character AI users spend 80 minutes daily talking to bots, hinting at AI companions becoming routine emotional outlets.
- Game devs embrace AI – 90% use it for testing, coding, world-building, but data ownership and privacy remain major concerns.
- Sam Altman revealed OpenAI’s $trillion future, including plans for data centers, new devices, and possibly acquiring Chrome.
🇪🇪 AI News from Estonia
- Estonian victims lost over €4M to scams in June-July, with criminals using AI-generated video calls and fake sites mimicking state agencies.
- Estonian startup BetterPic raised €2.5M to expand its B2B portrait AI tool, promising studio-quality images delivered in under an hour.
- TI-Hüpe AI education program launched with 8 team members and first teacher trainings last week, aiming to prepare schools for the AI era. Rewatch it on YouTube.
- AIRE (AI & Robotics Estonia) secured funding through 2028, aiming to double industry productivity and offer new AI support services.
- Biomed startup Elnora, founded by Estonian scientist Carmen Kivisild, PhD, moved to the US after failing to raise local investment. Read here.
- Rain Laane, head of Estonia’s Health Insurance Fund, revealed plans to use AI in family medicine to boost efficiency and patient overview.
- META Advisory’s Gregor Sibold highlighted how AI raises the strategic role of PR professionals in an era of algorithm-driven visibility. Read here.
🎙️ AIPowerment Podcast dropped a new episode featuring Sander Aavik, a B2B sales expert, discussing AI tools for automated email outreach.
🎧 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
