OpenAI launches a blog focused on the scientific use of AI
The company OpenAI has launched a new blog where researchers share findings, practical methods, and personal experiences using AI in scientific discovery.
Archive
November 9 – November 15
The company OpenAI has launched a new blog where researchers share findings, practical methods, and personal experiences using AI in scientific discovery.
A 2024 study compared the reasoning abilities of humans and AI language models on logic tasks. Older models performed worse than humans, while newer models have already surpassed them.
Author Sebastian Raschka published a technical tutorial on his blog Ahead of AI about positional embeddings in LLMs – from classic absolute and relative methods to RoPE, with math, visuals, and a PyTorch implementation.
Simon Willison released version 0.6a0 of the datasette-agent plugin. It brings minor fixes, an agent prompt directly on the Datasette homepage, and support for hash-mentions for tables and saved queries.
The datasette-llm plugin for the Datasette tool has been released in version 0.1a11 (alpha) with support for OpenAI-compatible API endpoints, making it possible to connect local models via LM Studio, Ollama, or llama.cpp.
The company Anthropic recommends that users of the Claude Opus 5.5 model abandon old prompts like "think hard", start at the Medium effort level, and specify the goal in a single message — according to the company, this speeds up responses without loss of quality and can also reduce costs.
According to researchers from Recorded Future and Halcyon, the number of ransomware attacks reported on leak sites rose to roughly 3000 in the summer of 2026, up 20–40% since spring. They say attackers are increasingly using LLMs to create phishing emails and negotiate with victims, but fully AI-driven ransomware has not yet emerged.
OpenAI is moving its robotics team into the core research division and the applied robotics division. The company believes the general-purpose GPT-6 Astra model handles robotic tasks better than specialized models.
The company Anthropic has reported, since 14 November 02:54 UTC, an increased occurrence of errors during the purchase, upgrade, and cancellation of plans on claude.ai. Core functionality of the Claude product is not affected; the cause is under investigation.
According to Reuters, Amazon is developing an internal project called World Product Graph – a database of products from across the entire internet, including goods that it does not offer itself. The goal is to keep users on Amazon and improve the AI agent Rufus (now Alexa for Shopping), which, according to an internal document, is often wrong.
GS Caltex connected no-code model development in Amazon SageMaker Canvas with production deployment in Amazon SageMaker Pipelines. According to the company, deployment time was shortened from a week to hours, and batch inference every 30 minutes reduced operating costs.
AWS has described the V2 platformVersion for Amazon Bedrock AgentCore runtime: instead of restarting the container, a session start restores a snapshot of the initialized environment, bringing session startup down to about 2 seconds, with pre-warming under 1 second.
GitHub Status reports reduced availability of the models gpt-5.6-sol-fast, gpt-5.6-sol, gpt-6.1-terra, and gpt-6-luna in Copilot products and IDE interfaces. The cause is an issue at an upstream model provider; choosing a different model or Auto mode is recommended.
According to an AWS case study, AssemblyAI runs thousands of GPU instances for speech-to-text through a combination of EC2 Spot and On-Demand instances, ECS, and SQS, reducing costs while maintaining SLA and data residency in European regions.
IEEE Spectrum highlights a gap in engineers' knowledge of AI auditing, ethics and governance, and according to IEEE, microcredentials make it possible to fill these competencies without interrupting one's career.
AWS Bedrock now allocates AI inference costs to individual IAM identities without code changes and links them to AWS Cost and Usage Reports. It also works for Claude Code, Claude Desktop, and OpenAI Codex, enabling per-user spending limits.
The GitHub Copilot app, from version 1.2.0, offers full support for GitHub Enterprise Server – working with remote repositories, issues, and pull requests. The feature was previously only in preview; now it is generally available.
Google describes the Gemini 4 Hydrogen model as faster, cheaper, and better in benchmarks than models from OpenAI and Anthropic. The model is now generally available in GitHub Copilot for agentic coding and multi-step tasks.
New: The model is also intended for document processing; Google presents the model as faster and cheaper than the competition; In benchmarks, it outperforms models from OpenAI and Anthropic; A new article from an independent source confirms the existence of the model
Next week the UN is meeting in Geneva to discuss a treaty regulating autonomous AI weapons. The impetus is deadly attacks – a drone killed soldiers near Bakhmut, and in July 2026 an AI drone killed three civilians in Zaporizhzhia. The author criticizes the current draft for gaps in accountability.
Google has launched the linking of Google Wallet with the Gemini product in the US: by connecting bank accounts via the Plaid service, users can track expenses, find forgotten subscriptions, and plan a budget. Access is read-only.
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