Archive
September 21 – September 27
ČVUT rector Michal Pěchouček said in a podcast that AI cannot be ignored, but the university must continue to teach algorithmic thinking, programming, and above all critical thinking.
GitHub has launched a public preview update to GitHub Copilot for Slack and Microsoft Teams with improved use of conversation context, model switching, and linking to GitHub issues. Available for Business and Enterprise plans, gradual rollout.
The AWS Machine Learning Blog described how to use the open-source framework SkyRL and the GRPO method to train the model Qwen3-VL-8B (Alibaba) on SageMaker HyperPod infrastructure. The success rate of solving the maze increased from 43.75% to more than 95%.
Meta's Muse app reached 3.4 million downloads in the first 17 days after launch (2.3–4.3 million according to other firms) and took first place in both the US App Store and Google Play. Its download growth rate significantly outpaces ChatGPT.
Using the example of its internal tool NarrateAI, AWS described five techniques for production quality assurance of LLMs on Amazon Bedrock – adaptive routing, multi-model failover, streamed evaluation, and data verification. The company states approximately 99% numerical accuracy.
AWS described the procedure for deploying the Qwen3-TTS-12Hz-1.7B-Base model (Alibaba Cloud) via SageMaker JumpStart to a managed endpoint. The model clones a voice from a short recording without retraining, supporting 10 languages including cross-lingual cloning.
Sony and Universal Music Group have filed another lawsuit against Suno. They claim that the v6 model infringes their copyrights because it was trained on the outputs of older Suno models, which were created from music illegally downloaded from YouTube and other sources.
Amazon Web Services and Qumulo have described a solution that, using Amazon SageMaker HyperPod and Cloud Data Fabric, enables training AI models with data located in a different AWS region without copying it. A test on the LLaMA v3 model at 60 ms latency showed performance comparable to that with local data.
The company OpenAI will end support for Custom GPTs in ChatGPT in December. Creators must switch to the new plugin system, but the migration will not transfer custom actions, conversation history, or sharing settings.
According to an analysis by The Conversation, a new category of work is emerging called AI remediation – people checking and correcting erroneous outputs from language models. White-collar workers in finance, tech and retail report that this is actually adding to their workload.
The FAA is deploying the AI system SMART from Air Space Intelligence for 875 million dollars over 12 years to predict and prevent air traffic delays. Launch in the Washington D.C. area, with controllers remaining responsible for safety.
{"text_5": "A fault in a submarine cable between China and Japan has reduced network capacity in Asia-Pacific, which, according to Hugging Face, is slowing down downloads of models and data from its platform in this region."}
According to the E-commerce Study 2026 research by the APEK association, 50% of Czech customers have used AI when shopping online, most often for comparing products, summarizing reviews, and searching for alternatives. So far, people use AI agents only to a small extent.
The government pilot program WISeR, which has used AI to approve or deny care in US Medicare since January 2026, has been plagued by technical problems, delayed decisions, and disputed denials. The GAO found a procedural violation in the program's implementation, yet it continues and an expansion is being prepared.
Developer Cresco Real Estate has launched the AI assistant Yarda on the website of the Yards Žižkov project, which filters apartments based on natural-language requests and answers questions 24/7. This is a pilot test of acceptance of AI communication among Czech clients.
According to a report, the White House asked OpenAI and Anthropic to have new AI models reviewed by a US agency first, before providing them to the British AI Safety Institute. Anthropic has already complied and made Claude Mythos 5.1 available only to US organizations.
Attorney Barbora Vlachová of the Czech Bar Association warns that AI results cannot be taken as the sole truth, and cautions against uploading contracts with personal and factual data into AI tools.
OpenAI has fixed a bug in image encoding in the GPT-6 Sol and GPT-6 Luna models that was degrading visual input understanding in both the API and the Codex product. The company recommends re-verifying affected workflows.
Independent commentator Simon Willison claims that the more he works with coding agents, the more convinced he is that they make software engineering harder. According to him, their full potential is unlocked only by exceptional discipline and knowledge.
PrismML has introduced the 1-bit Bonsai LLM with 2 billion parameters for smart glasses on Qualcomm's Snapdragon AR1 Gen 1 platform — the model runs locally without sending data, though no specific glasses using it have been announced yet.
AI Radar monitors Czech and international sources every day, looking for changes that truly deserve attention.
The result is a quick overview of what has actually changed in the AI world, rather than another stream of articles.
Each row represents one event — not one article. At a glance, you can see its significance, credibility and main point.
Agent mode is available to all paying users
Until now, the mode was available only on the highest plan; it is now available on all paid tiers without a waitlist.
When you open an event, you will find a more detailed summary, an explanation of its significance, and links to the original sources. Quick updates are available separately in AI Flash; important topics move to Today’s Overview once more information is added.
We first summarize each event in the same way for everyone. Based on those same facts, we then explain what the change means for your own use and what it could mean for how a company operates.
Today’s briefing is the same for everyone. Pages For individuals and For businesses can be found in the main navigation — they select only events relevant to the given use case.
