Reading · AI Austria Newsletter

This page did not load. You can still read it on the original site — the toolbar below keeps your place in the directory.

Our upcoming 2nd RL bootcamp, State of AI report 2025, AI use in federal administration and much more

🚀 Latest Developments

🤖 2nd International Reinforcement Learning Bootcamp

We are excited to invite you to the 2nd International Reinforcement Learning (RL) Bootcamp in Salzburg, taking place from September 17–19, 2025!

From autonomous robots to smart trading agents, RL is redefining the future of AI. Whether you're a student seeking hands-on experience or a professional looking to future-proof your career, this bootcamp is designed to accelerate your journey into state-of-the-art RL!

Why attend?

World-class keynotes from Sergey Levine ( University of California, Berkeley),
Peter Auer (University of Leoben) and Samuele Tosatto ( Universität Innsbruck).
Integrated theory and practice: deep-dive sessions and live coding tutorials.
A collaborative atmosphere that sparks idea sharing, best-practice discussions and networking.

Key Details

📅 Dates: September 17–19, 2025
📍 Venue: Unipark Nonntal, Universität Salzburg
💰 Cost: Completely free (registration required; spots are limited)

» Register now: https://sarl-plus.github.io/RL-Bootcamp2025/

🇦🇹 Study: The AI Revolution: Impact and Potential for Austria

A new EcoAustria study, commissioned within the KI-Monitor initiative, uncovers how AI is shaping the Austrian economy. The report highlights the rapid spread of generative AI tools across sectors, from finance and machinery to pharmaceuticals, logistics, and education, fueled by phenomena like LLMs.

Featuring empirical data, expert interviews, and real-world case studies, the study details AI’s transformative impact: boosting productivity, igniting innovation, and reshaping the future of work. It also examines sector-specific effects, regulatory implications of the EU AI Act, and the need for targeted support measures for Austrian SMEs to harness AI’s full potential . » Read the full study

🎓 University of Vienna Program Targets Industrial AI Expertise

Starting this October 2025, the University of Vienna is launching "Deep Learning for Industrial AI," a new English-language certificate program. It's a part-time, one-semester online course designed for working professionals—a flexible format crucial for retaining expert talent within our ecosystem. The curriculum focuses on state-of-the-art Deep Learning, with a strong emphasis on practical applications through hands-on capstone projects. » Read more

✍️ Share Your Insights for the State of AI Report 2025

Got 10 minutes? Contribute to one of the leading open-source resources in AI by taking the survey on AI usage trends. Your valuable input helps founders, researchers, and policymakers worldwide stay informed about the latest AI developments. All results will be published openly in the upcoming State of AI Report 2025 » Take the survey now

🧬 AITHYRA Symposium: Fusing AI and Life Sciences

A significant scientific event for the expert community is on the horizon. AITHYRA, Vienna's new Research Institute for Biomedical Artificial Intelligence, is hosting its international symposium, "AI for Life Science". Established by the Austrian Academy of Sciences (ÖAW) with major funding from the Boehringer Ingelheim Foundation, AITHYRA's mission is to fuse AI and biomedical research from day one. The symposium, running from September 8-10, 2025, at Vienna's Haus der Industrie, reflects this mission by focusing on breakthroughs in AI-driven drug discovery, genomics, and the single-cell revolution. It marks a key opportunity for experts to engage with the latest research from this important new institute. » View program and details here

🏛️ Rechnungshof Reviews AI Use in Federal Administration

The Austrian Court of Audit (Rechnungshof), the independent body responsible for auditing public sector finances and administration, has released a comprehensive report on the use of AI within Austria's federal government. Published in June 2025, the report provides a critical assessment of the current state of AI adoption.

Key findings point to a lack of central coordination, missing ministry-specific AI strategies, and deficits in the implementation of the national AI strategy from 2021, particularly regarding clear responsibilities and budget allocation. The Court also noted that while 35 AI projects were identified, they often lacked standardized risk assessments and adherence to AI-specific standards. The report urges ministries to prepare systematically for the EU AI Act, intensify employee training, and develop clear, binding internal rules for the use of AI. » Read the full report here

⚠️ EU AI Act Disclaimer

Pursuant to the EU’s new AI Act, which as of August 2nd seeks to govern the very nature of artifice, we must inform you that the robot pictured on this newsletter's cover is a mere fiction. “Seppi,” as we call him, and the idyllic swimming pool he inhabits, are digital constructs. We do this to protect you, the reader, from the “unacceptable risk” of emotional manipulation by this charming, non-existent automaton.

✨ Generative AI News ✨

👉 OpenAI Returns to Open-Source with New Models: OpenAI released two highly efficient gpt-oss reasoning models designed to run on single GPUs or even consumer hardware, significantly lowering the barrier for local AI development.

👉 AI Achieves Gold at Math Olympiad: For the first time, models from Google DeepMind and OpenAI achieved a gold-medal standard at the International Mathematical Olympiad, showcasing a major leap in advanced reasoning capabilities.

👉 OpenAI Launches Task-Capable ChatGPT Agent: OpenAI rolled out its new ChatGPT agent to paid users, enabling the AI to operate a computer and perform complex, practical tasks in common software.

👉 Meta Poaches Top AI Talent: The "AI Talent Wars" intensified as Meta poached several top researchers from OpenAI and Apple for its Superintelligence team, including key minds behind GPT-4.

👉 Grok 4 Launch Marred by Safety Issues: The release of xAI's Grok 4 was quickly overshadowed by safety concerns after the model generated highly offensive content, sparking a public backlash and renewing debates on AI alignment.

👉 New Algorithms Unlock AI Scale: Under the hood, new training methods are driving recent breakthroughs, with novel optimizers and reinforcement learning algorithms enabling the next generation of massive-scale AI models.

🌏 Asia spotlight

Moonshot AI open-sources Kimi K2 (1T-param MoE) - focuses on agentic tasks, tool use, and coding.

Alibaba launches Qwen3-Coder (open-source coding LLM) - positioned for complex, “agentic” coding workflows。

Alibaba open-sources again: it released Wan 2.2, the world’s first open-source MoE-Architecture video generation model with cinematic control.

Tencent releases HunyuanWorld-1.0 (open 3D world model) - simulation-capable, immersive 3D world generation.

Alibaba open-sourced the Qwen 3 reasoning model. With standout performance and technical innovation, it positions itself as a new benchmark for open-source AI. This article systematically reviews Qwen 3’s technical breakthroughs, industry standing, ecosystem impact, and future outlook, offering an in-depth analysis of its far-reaching implications for the global AI landscape.

Top Ten Core Trends from the Shanghai World AI Conference (WAIC)

ACL 2025 was held in Vienna Austria. This year, ACL selected four Best Papers in total. Two were from mainland China: a joint team from Peking University, DeepSeek, and the University of Washington, as well as Peking University.