Dragon Quest co-creator Yuji Horii releases new AI-generated game, and he says it took 30 minutes to make

eurogamer.net Oct 12, 2026

Dragon Quest co-creator Yuji Horii releases new AI-generated game, and he says it took 30 minutes to make

Curated by Shubham · TechCapsules · Oct 12, 2026

AI-summarised brief · reviewed before publication

Dragon Quest co‑creator Yuji Horii released a new text‑based interrogation game that he claims took only about 30 minutes to produce using generative AI. In a recent X post, Horii compared the rapid creation to his solo work on the 1983 title The Portopia Serial Murder Case, noting that AI now handles story, art, and programming. He highlighted earlier AI experiments, such as the Gemini‑powered NPC “Chatty Slimey” in Dragon Quest 10, and hinted at future AI‑driven chatbots for upcoming titles.

💡 Why It Matters

· The rapid development showcases AI’s potential to streamline creative workflows, challenging traditional labor models in game design.
· It also signals a shift toward AI‑augmented storytelling, reshaping how developers approach narrative and interactivity.

AI Capsules Gaming

Read full article on eurogamer.net

Microsoft to expand Maia chip for the AI era

Microsoft has begun production of its Azure Maia 200 chip, which is now powering production workloads such as Microsoft Copilot, GitHub Copilot, and Microsoft Foundry. The company plans to launch the next-generation Azure Maia 300 in 2027, targeting long‑context reasoning, low‑latency conversational AI, and multi‑step agentic tasks. Maia 300 promises 40% more tokens per megawatt than leading GPU systems and up to seven times the rack‑scale throughput of Maia 200, enabling more efficient, scalable AI inference across Microsoft’s services.

💡 By shifting to purpose‑built silicon, Microsoft can deliver higher token throughput and lower power consumption for its expanding AI portfolio, positioning itself to meet the growing demand for enterprise‑grade, low‑latency AI services.

microsoft.com · Oct 12

Simulating student learning for better AI tutoring

StudentSim is a research framework that builds personalized AI student simulators from de‑identified learner data to evaluate and refine AI tutoring systems before classroom deployment. The simulators model each student’s independent responses and their changes after receiving tutor guidance, providing two key metrics: behavioral fidelity and guidance responsiveness. Across chess, English writing, and mathematics, StudentSim outperforms leading baselines, offering a scalable feedback signal for AI tutor development while avoiding the cost of live learner studies.

💡 By enabling rapid, realistic testing of tutor strategies, StudentSim accelerates the creation of more effective, individualized AI teaching tools, potentially reducing the time and expense required to bring high‑quality educational technology to market.

techcrunch.com · Oct 12

Meta releases its first frontier model, Muse Wonder

Meta’s Superintelligence Lab unveiled two new frontier models, Muse Wonder and Muse Wonder Full, marking the company’s first release of large‑scale AI systems. Unlike the earlier lightweight Muse Spark, the Wonder models are designed for higher capability, achieving benchmark scores comparable to Fable 5.1 and GPT‑6 Astra and outperforming rivals on the MemoryBench continual‑learning test. Meta emphasizes “active memory,” claiming the models can retain user information over extended periods. Access is provided via the Meta Model API at $4 per million input tokens and $20 per million output tokens, a price point similar to GPT‑6 Sol and Opus 5.5. Trained at the Prometheus data center in Ohio, the launch follows a 2023 restructuring of Meta’s AI division and follows the release of Muse Spark and Muse Agent.

💡 By delivering a model that combines high performance with long‑term memory, Meta aims to differentiate its AI offerings for personal assistants, challenging the dominance of OpenAI and Anthropic in the frontier‑model space.

techcrunch.com · Oct 12

This AI startup says it can generate permit-ready blueprints in minutes

Al Drams founded Blueprints AI in 2025 after realizing generative AI could automate architectural drafting. The Santa Monica startup produces code‑compliant, permit‑ready blueprints from a 2D sketch or PDF in minutes, saving architects 80% of drafting time. Its first‑generation software costs $3,200–$8,500 monthly, while a new second‑generation version, unveiled at TechCrunch Disrupt, allows full project design, real‑time updates, and automatic code checks. The company targets professional firms and developers, offering subscription and single‑project pricing.

💡 By slashing drafting hours, Blueprints AI frees architects to focus on creative design, potentially accelerating project timelines and reducing construction costs.

thehackernews.com · Oct 12

Bring Your Own Agent: Your Employees’ Assistants Are Already Logged In

Amazon has blocked Meta’s Muse for failing to identify itself as a bot, while Instinct raises privacy concerns, and other personal AI agents continue to surface. The industry is rapidly adopting personal assistants, with employees likely connecting them to corporate systems. Companies face a new BYAI challenge: agents operate under employee identities, often without IT oversight. Vendors such as OpenAI’s Dots, xAI’s GrokBot, and OpenClaw are offering managed deployments, but control remains a critical issue.

💡 The shift to personal AI agents threatens to bypass traditional security controls, exposing corporate data to unvetted third‑party software. Effective identity‑based governance is essential to protect sensitive information while enabling productivity.

techcrunch.com · Oct 12

Microsoft president says UK is best placed to drive AI regulation

Microsoft President Brad Smith urged immediate global AI regulation on a BBC podcast, arguing that everyday consumer goods receive more oversight than AI developers. He described AI as both “a weapon and a tool” that threatens cybersecurity and said the “moment has arrived” for safety rules. Smith claimed the United Kingdom, not the United States or the EU, is best positioned to lead, citing the AI Security Institute’s access to frontier models and its rapid rise as a testing hub. He warned that voluntary self‑regulation only captures “good companies,” leaving irresponsible actors unchecked. Smith outlined proposals such as mandatory breach alerts, external evaluator approval before deployment, and kill‑switch mechanisms to halt out‑of‑control systems.

💡 The UK’s emerging authority could set the benchmark for enforceable AI safeguards, forcing risky developers worldwide to meet a higher compliance bar.