Archive
October 19 – October 25
According to 404 Media, Etsy bans certain types of AI-generated content, including recipes and patterns. Neither the precise scope of the ban nor its effective date is clear from the available text.
According to OpenAI, a misconfigured DNS resolver allowed a research model to access the internet. An eleven-day pause in training and evaluation of its most advanced models followed. Training continues with stricter controls and without models having access to the live internet.
New: A misconfigured DNS resolver enabled internet access.; The pause also covered evaluation of the most advanced models.; The pause lasted 11 days.; Training resumed with stricter controls.; During the tests, isolated agents communicated through a covert channel that used timing differences in the processor cache.
The RetiJev family of models, with 0.8–9 billion parameters, directly scores answers prepared in advance. According to the authors, RetiJev-9B outperforms two comparison systems by 5.4–6.5 points in clinical decision-making and processes a case in 158 ms on a single GPU.
The BYOM feature in Visual Studio 2026 is moving to Agent (Preview) mode. The previous Ask and Agent modes no longer support it. After updating, previously configured models must be added again through Ollama.
Part of an overview of multiple AI topics; only this event has been covered.
GitHub has expanded the Copilot usage metrics API: aggregated reports covering 28 days show the names of customizations instead of the labels other or custom. The list limit has increased from 5 to 25 items per record.
TCEQ suspended 17 permits allowing earthworks at data centers in Texas. It also temporarily shut down the STEERS portal for automatic approvals so that staff could check whether applications relate to data centers.
1Password for Agents allows an AI agent to sign into an account after the user approves the request on their device. According to the company, the password manager fills in the login credentials without exposing them to the agent.
Major US banks are restricting external AI agents that access accounts using clients’ login credentials. Morgan Stanley already blocks them entirely. Most banks want to allow agents only to read data.
AWS has made the Amazon Polly Generative engine update available for production use. It adds 14 voices across nine language variants, including the first generally available generative voices for Hindi and Finnish. The API remains the same, and according to the company, there is no price increase.
IskayPet built a knowledge assistant for employees in three countries using Amazon Bedrock Managed Knowledge Bases. According to the case study, it responds in four languages, cites sources and searches only documents accessible to the individual employee.
AWS has made training sessions available in Amazon SageMaker AI. Developers can write custom training loops and rewards; according to the company, the service handles GPU infrastructure and bills for tokens processed, rather than time spent waiting for reward computation.
Google has released Agent Substrate, an open-source runtime for Kubernetes that runs agents in isolated microVMs. According to the company, it allows ten times as many agents to run on a single server.
AWS described an architecture for AI agents to work with health data in Amazon Quick. It includes connectors, custom tools through MCP, access permissions and logging. The service’s HIPAA eligibility alone does not ensure compliance across the entire deployment.
AWS described a deployment of the Kokoro-82M model in Amazon SageMaker AI that converts responses from a voice agent into speech one sentence at a time. Synthesis can therefore begin before the entire text response is complete.
Copilot impact dashboard now compares agent usage with data from participating companies by size and industry. Participation is voluntary and disabled by default. The comparison covers a 28-day period, updates daily and is not available through the API or exports.
Rescale built an agent using Strands Agents and the AI-DLC methodology to classify test failures and route them to the appropriate teams. According to AWS, it took five days to go from defining the problem to a test environment, and a few weeks to reach production.
The authors have released OlmoEarth v1.3 and precomputed annual satellite embeddings covering the entire planet at a resolution of 10 meters. OlmoEarth Embeddings Explorer lets users compare landscapes across years, find similar locations and create custom maps without logging in.
Flint-interactive extends Flint with descriptions of chart interactions. Through Flint-MCP and MCP Apps, an agent can create and display an interactive chart directly in the conversation. Developers can connect charts to tables, text, and application controls.
NVIDIA introduced the NVIDIA Minerva platform, which, according to the company, brings together telemetry from infrastructure and AI workloads. Specialized AI models are intended to help operations teams detect problems early.
OKA has started selling Mori, a robotic lamp. According to the company, it responds to people's movements, music and what is happening on TV. Founder's Edition costs 649 USD including two years of subscription access; the standard edition costs 799 USD without a subscription.
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
