Archive
September 14 – September 20
In the State of the Union 2026 address, European Commission President Ursula von der Leyen warned of the risks posed by AI agents and self-improving models and announced a plan to cooperate with Canada and the United Kingdom on model safety, citing AI Act as a tool for safeguards.
A journalist at WIRED created PitchFly, a project that uses the freely available drosophila brain connectome (166 000 neurons, 125 million synapses) to generate ideas for article headlines – without understanding the meaning of words, simply by remixing learned patterns.
AWS has released an open-source collection of 38 agent skills for 11 areas of healthcare and life sciences. According to the company, AI agents with these skills win 70–86 % of head-to-head comparisons against agents without them. MIT-0 license.
AI Now Institute and Local Progress are seeking a researcher to create three state-specific guides to local data center regulation, building on the organization's North Star Data Center Toolkit.
According to The Information, Apple is developing an enterprise server for AI inference with two or four M8 Ultra chips, potentially with NVLink Fusion technology from Nvidia. The project was backed by current CEO John Ternus, and a launch is not planned before 2029.
New: The server exists in two versions: with 2 or 4 M8 Ultra processors; Focused on AI inference (running trained models, not training them); John Ternus (the new CEO of Apple) backed the project about a year ago while leading the hardware division; OpenAI and Anthropic are already buying Mac Mini and Mac Studio for AI workloads, signaling demand; The project may have recently accelerated
Google launched early access to an MCP server for Google Home – AI agents such as Claude, ChatGPT or Google Antigravity can control smart devices through natural language. Currently only for Google Home Premium Advanced subscribers in the USA.
Anthropic has merged Claude chat and Cowork into a single interface and added features for creating and sharing presentations and documents. Initially for Pro and Max plans, with free and team plans to follow later.
Mustafa Suleyman warns that AI models should not be attributed consciousness, feelings or rights – doing so would, in his view, make AI safety and alignment more difficult.
Zoom introduced the AI tool Revenue OS with Engage and Forecast features, built on the acquisition of Common Room in July 2026, with the aim of competing with CRM platforms from Salesforce and HubSpot.
Amazon Bedrock AgentCore now offers optimization of agent system prompts: it uses production traces to suggest a revised prompt with an explanation, which is validated through offline evaluation and A/B testing before deployment.
TypeSafe AI clarified the price of Jev (0.042 USD/million input tokens, output free), and commentators point to the risk of hidden bias and the absence of explanations for decisions in this classification model with no text output.
New: Model price: $0.042 per million input tokens, output free; Risk of hidden bias in the classification model; The model operates as a black box without explaining its decisions; Applicable to spam filtering and search reranking; TypeSafe AI categorizes it as 'System One models'
AWS described a serverless pipeline combining Amazon Bedrock Data Automation, Step Functions and Lambda for automatic PII detection and redaction in documents using foundation models and custom blueprints without training custom ML models.
According to its announcement, NVIDIA achieved leading performance with the Vera Rubin NVL72 platform in its debut in the MLPerf Inference v6.1 benchmark. The source does not provide specific figures.
Anthropic drew attention to a network of around 28 dating apps (Dora, Romi, Doni), where conversations were conducted predominantly by autonomous AI personas powered by the model Claude. Only 1 in 4 profiles belonged to a real, paid worker; the apps were still in US app stores.
OpenAI has added the Czech startup Deepnote to the native tools in ChatGPT Work. An AI agent can thus create and run notebooks in Deepnote and analyze company data without exporting it from the platform.
Seznam has launched a language model on Zboží.cz that allows users to dictate a purchase review instead of writing it. AI extracts pros, cons and a summary from speech and fills in the form; the text can then be edited by typing or dictating.
GitHub has enabled AI Scan to find security issues in pull requests even without CodeQL default setup configured. The feature is in public preview for GitHub Advanced Security customers on github.com; GitHub Enterprise Server is not yet supported.
Syensqo says it uses AI agents to digitally synthesize millions of molecular combinations and predict their properties in order to accelerate the development of materials for data centers, cooling and semiconductor manufacturing.
OpenAI reported and subsequently resolved an incident involving elevated error rates for the model gpt-image-2.5-flare in the Images service; according to the company, all affected services have fully recovered.
A poll by New York Times and Siena University among 1 503 likely voters (early September 2026) found that 61 % oppose the construction of data centers for AI and only 14 % strongly support it. Views differed only slightly between voters who supported Trump and those who supported Harris.
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.
Use the CS/EN switch to read the same Radar in Czech or English. English content is published after its translation has been checked, so new and older items may appear later.
Each row represents one event — not one article. At a glance, you can see its significance, credibility and main point.
Illustrative example, not a current news item.
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.
The detail page contains a fuller summary, its significance and original sources. Practical impact appears in the detail and the For individuals and For businesses views. An AI Flash item reaches the main selection only after it has been expanded and meets the publication rules.
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.
Reset history · Outage history · Subscribe in a feed reader (RSS) · Methodology and label explanations
