AI Flash · live feed
A timeline of everything that has passed through the radar — newest first, with an evidence status for each item.
Continuously updated
1146 items in the selected feed
How to read evidence statuses ↓
Startup AegisAI, founded by former Google security managers, has raised 36 million dollars in a Series A round led by Battery Ventures to develop AI agents to combat targeted phishing attacks (spear phishing).
Patreon is laying off 93 employees (20 % of the team). CEO Jack Conte explains the decision by saying that AI is fundamentally transforming the technology industry and that companies that do not fully adopt AI tools will not be competitive in three years.
Jefferies built an AI agent-based trading assistant on AWS that provides traders with real-time analysis of data from multiple sources using natural language, without coding. The solution uses Anthropic's Claude, Amazon Bedrock, Knowledge Bases and MCP to connect to market data. The assistant accelerated data exploration from days…
Amazon Bedrock AgentCore is an optimization feature that detects silent AI agent failures — errors without error signals, such as incorrectly performed actions or omitted checks. The tool analyzes traces using a taxonomy of 11 failure categories and groups them into patterns. According to Amazon, it helps distinguish…
AWS is expanding Amazon Bedrock Managed Knowledge Bases with agentic retrieval mode — the new AgenticRetrieveStream API automates multipart queries instead of performing a single search. The model breaks complex questions into subquestions and searches iteratively, addressing the shortcomings of traditional top-k…
Meta launched an advertising campaign promoting AI as a positive force, accompanied by David Bowie's song "Five Years". The irony: the original song is about humanity's panicked response to news of mass extinction in five years.
Etched closed a 300 million dollar Series C funding round at a 10.3 billion dollar valuation, led by Sequoia and Andreessen Horowitz. The startup doubled its valuation in 7 months. It developed AI inference chips optimized for the prefill and decode phases of transformers and offers them as complete systems with 1…
Startup Lunar Outpost plans to use Nvidia Jetson chips on a lunar rover controlling a lidar system — probably the first GPU on the lunar surface. Nvidia is also working with Firefly Aerospace to deploy Jetson on a satellite orbiting the moon for image processing. Jetson is compact and energy…
Two US lawmakers are preparing to introduce a bill that would require AI companies to have the ability to shut down or slow down their systems on orders from the Department of Defense. A shutdown would be triggered in scenarios involving the deaths of at least 10 people, economic damage exceeding 100 million dollars…
NASA JPL and Loft Orbital tested the Gemma 3 model from Google directly aboard the YAM-9 satellite — the NAVI-Orbital system analyzes images on-board and sends a text summary instead of raw data, with 88% accuracy in a ground-based test.
AI accelerates the development of biologic drugs by computationally generating and filtering candidate molecules. AstraZeneca uses a build-measure-learn cycle to shorten development time. It has potential to address targets previously considered out of reach.
Google raised its investment forecast for 2026 to 195–205 billion dollars. Q2 revenue reached 119.8 billion USD (+24 %), the Gemini app has 950 million MAU, and AI Mode in Search exceeded 1 billion. Google is training Gemini 4 with larger foundation models to compete, focusing on improving coding and agentic…
The US administration claims that Moonshot copied Fable from Anthropic, but experts Braden Hancock and Nathan Lambert are skeptical. They say LLM distillation alone cannot create such an advanced model in two weeks (Fable has been publicly available since 1 July). Advanced techniques such as reinforcement learning would be needed.
Startup Lingverse raised 29 million dollars to develop iKairos — a wearable AI device that can be worn as a pendant or clipped to clothing. It is intended to serve as a personal AI assistant with voice commands and task offloading, priced at 199–249 dollars with a possible subscription service. Lingverse claims patents from…
ServiceNow is investing 40 million dollars in BusinessNext, acquiring an approximately 5% stake at a valuation of 700 million dollars. BusinessNext, with revenue of 32 million dollars, develops a platform with AI agents to automate banking workflows; its customers include the Indian central bank and the largest banks. The partnership will combine ServiceNow's workflow systems with BusinessNext banking expertise to sell AI solutions to the financial sector.
MAGE is a multi-agent framework for optimizing macro placement in chip design. It combines floorplanning rules, visual checks and iterative improvements using natural language directives rather than learning from data. On designs using NanGate45 and GlobalFoundries technologies…
A research team examines the security of the "alignment layer" in graph foundation models by testing 6 public systems. Direct perturbation in representation space was able to collapse all models; an accessible input attack halved accuracy in 3 of the 6 tested systems. OpenGraph was the weakest…
A scientific position paper on arXiv criticizes deepfake research for focusing on epistemic harms (truthfulness, authenticity) while ignoring practical subject-centric harms — especially AI-generated nonconsensual sexual content (AIG-NCII). The authors recommend reframing threats and integrating safety…
NVIDIA brought the DGX GB300 supercomputer to the US military academy in Monterey. The system gives 1500 students and 600 professors access to train large models and run inference for applications in weather prediction, cybersecurity and resilience planning.
A technical analysis of Nvidia's new Vera Rubin NVL72 inference hardware. According to early results from CoreWeave, Vera Rubin delivers 5.4× higher performance per MW and a 5× better performance/price ratio than GB200 NVL72. The hardware is in early development; Nvidia plans to publish verifiable results by Q3 2026. The figures…
A service disruption incident involving Haiku 4.5 affected Claude.ai, Claude Console, Claude API, Claude Code and Claude Cowork. The cause was identified, a fix was deployed and the services were monitored. The incident was resolved.
Security researcher Thomas Ptacek claims that an open LLM model from 2025 with a purpose-built penetration testing tool could carry out a sandbox escape and scan or hack most corporate networks. He says this is unsurprising and merely confirms weaker security than at OpenAI.
IBM reported revenue of 17.2 billion dollars in the latest quarter, with mainframe sales down 42 %. Management says customers decided against upgrading mainframes because of high prices for AI hardware and data center equipment, to which they shifted their budgets. The CEO promises a return to normal.
The White House and the US administration's Commerce Department disagree on the approach to Chinese AI models. The White House wants strict controls against distillation attacks (training models based on US systems), while the Commerce Department considers this impracticable. The reason is concern about the model…
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
