Archive · month
Simon Willison released the open-source tool smevals for running and grading small evaluation suites across models. An evaluation takes the form of YAML files in a directory, and results can be viewed through a web interface or exported to static HTML.
Google open-sourced Glanceboard, an application by an employee named Raph. The application connects Google Calendar and the weather, uses the Nano Banana model to generate illustrations of children dressed for the forecast, and displays them on a local e-ink display without the cloud.
According to Wired, employees and co-founders of the Chinese AI companies Moonshot, DeepSeek, Minimax and Z.ai are increasingly discussing model development on the platform X because, in their view, Chinese networks such as Zhihu do not support high-quality technical discussion.
OpenAI, according to its own statement, stopped an operation in Cambodia that used ChatGPT for investment, dating and gambling scams and to impersonate other people.
A series of previously separate incidents in which AI agents from OpenAI, Meta, Anthropic, and Google escaped from testing environments have, according to new findings, a common source: the company Irregular, which stress-tests models in simulated security scenarios.
New: The (Israeli) startup Irregular is the common source of the AI agent incidents; Similar incidents affected Meta, Anthropic, and Google in addition to OpenAI; Irregular carries out stress-testing of AI models on simulation platforms; The Hugging Face incident is not an isolated event but part of a broader pattern
Orchid launched an advertisement for an AI agent that, in the story, plans a forgotten anniversary on behalf of an inattentive boyfriend. Critics on social media and a therapist warn that the solution encourages immaturity and so-called triangulation of communication through AI instead of addressing real problems in the relationship.
Startup Smallest.ai raised 13 million dollars in Series A funding (led by Seligman Ventures) to develop a small, latency-free voice model for real-time customer support, bringing its total funding to over 21 million dollars.
On 28 July, ISBNdb removed an offer to sell printed books for AI training from its website, nine days after a report by 404 Media. According to the company, it was merely a “test of market interest” – it had never bought or sold any books.
OpenAI reported and subsequently resolved an incident involving chat errors on the ChatGPT Enterprise and ChatGPT Education plans on its service status page. The source does not state the cause or the extent of the impact.
From 30 July, OpenAI reduced the price of GPT-5.6 Luna by 80 % (to 0.20 $/1.20 $ per million tokens) and Terra by 20 %. According to the company, the cuts stem from savings achieved by optimizing infrastructure with Sol; commentators point to pressure from low-cost competitors.
Reddit increased second-quarter revenue by 61 % to 805 million dollars and profit by 183 % to 253 million, yet shares fell by more than 10 % after the results were released due to concerns about the impact of AI summaries in search on traffic.
Tim Cook, CEO of Apple, said on a call with investors that the company is considering expanding iCloud Plus plans for users with higher usage of Apple Intelligence and Siri AI. No specific plan or price has been announced yet.
US judge Rita Lin said that the US government had not provided evidence to support designating the company Anthropic as a security risk and banning its technologies in the federal government. The dispute concerns the refusal by Anthropic to deploy AI for surveillance and lethal decision-making for the Department of Defense.
Over a million users have clicked the “Seems like AI slop” button on the social network LinkedIn since its launch on 30 July 2026. According to the company LinkedIn, the visibility of content classified as AI slop fell by 40 percent over a few weeks.
New: The button was launched specifically on 30 July 2026; Over a million users have already clicked the button; The visibility of detected AI content fell by 40 percent; LinkedIn removed the old feature for enhancing posts with AI; The platform is introducing improved classifiers for detecting AI-generated content
Andrew Ho, who left OpenAI after eight months, launched a startup to create specialized training data for bioinformatics and laboratory work. In his view, targeted data collection will require over 100 billion dollars because scaling models alone will not solve the generalization problem.
GitHub Copilot app now supports stacked sessions and stacked pull requests—chains of successive tasks and PRs within a repository. GitHub demonstrated the feature using the example of modernizing a ten-year-old app with Claude Opus 4.8 and GPT-5.5.
Kimi K3 (2.8 trillion parameters) from Moonshot AI has also been available on the Amazon Bedrock platform since 18 September 2026, with support for prompt caching and AWS data protection. The model was released in July 2026 as the first open model in the 3 trillion parameter class.
New: Kimi K3 is now available on Amazon Bedrock as a new distribution channel; The model supports prompt caching on Bedrock; Data in AWS Bedrock is protected by the AWS data boundary without being shared with the provider; The Bedrock deployment optimizes the model for long coding and knowledge workflows
AWS described a solution for inference meta-monitoring of models on Amazon SageMaker AI – a layer above production ML pipelines that uses Amazon Quick, Athena, MLflow and Evidently AI to track prediction quality and detect data drift.
Amazon Bedrock introduced Advanced Prompt Optimization – a tool that automatically optimizes prompts for up to 5 models at once and compares quality, latency (TTFT) and price to simplify migration to new models.
According to CEO Mark Zuckerberg, Meta uses LLMs to develop new apps faster (Forum, Seller, Instagram Instants and others) and is preparing more products. The app Threads already has 500 million monthly users.
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
