Today's briefing
- Agents
Docker launches Docker Agent for defining AI agents in YAML
AgentsGeekNewsOct 8, 2026 - Research
OpenAI shows proofs of 372 mathematical results including Unique Games Conjecture
ResearchGeekNewsOct 8, 2026 - Physical AI
Boston Dynamics names Rohit Prasad as new CEO
Physical AIGoogle News +5 moreOct 8, 2026
Korea Desk
What Korea’s tech press is reporting, summarized in English.
9 stories in this issue come from Korean outlets; those outside the desk carry a KOREA mark. Originals are in Korean.
- Physical AI
TwelveLabs launches Pegasus 1.6 for robot learning from first-person video
TwelveLabs released Pegasus 1.6, a vision-language model specialized for collecting physical AI training data from first-person video, automating understanding and labeling of egocentric footage.
AI TimesOct 7, 2026 - Industry & Policy
Anthropic expands Project Glasswing to reach more security professionals
Anthropic expanded its Project Glasswing program to give more cybersecurity professionals access to Claude models through a new tiered system with Defense, Red Team, and Specialized levels.
AI TimesOct 7, 2026 - LLM
Mistral AI releases Le Chonk, a 1 trillion parameter multimodal model
Mistral AI unveiled Le Chonk, a natively multimodal model with 1 trillion parameters and 49 billion active parameters. The model is released as open-weight, marking a major update from the European AI company.
Byline Network +1 moreOct 7, 2026 - Agents
Google explores expanding Gemini's call-making feature to personal contacts
Google is developing the ability for Gemini to make calls to personal contacts including friends and family, beyond its current business-focused calling feature. The expansion would allow AI to handle simple tasks like relaying messages or confirming plans.
AI TimesOct 8, 2026
Stories in this issue
34 storiesChatGPT's Intelligent UI combines text responses with interactive visuals
OpenAI launched an Intelligent UI feature in ChatGPT that allows the chatbot to answer questions with interactive visuals. Rolling out alongside GPT-6, the update gives ChatGPT the ability to combine text responses with diagrams, charts, forms, and tappable buttons.
DeepSeek nearly doubles pre-IPO funding to up to $15 billion
DeepSeek raised its pre-IPO funding target from $7.4 billion to between $12 billion and $15 billion as investor interest surged. CATL and Tencent Holdings are the lead investors.
Mistral AI unveils Mistral Large 4, a 1 trillion-parameter open model
Mistral AI released Mistral Large 4, an open-weight model with over 1 trillion parameters, currently in public preview via API before the model weights launch on the 27th.
OpenAI publishes 722 papers solving 90 of the top 500 open math problems
OpenAI published 722 papers that solve 90 of the top 500 open mathematical problems, marking what is being called the most significant moment in over 100 years of mathematics.
Docker launches Docker Agent for defining AI agents in YAML
Docker released Docker Agent, a CLI plugin that defines and runs AI agents in YAML without writing code, supporting multi-agent teams for automated task delegation and collaborative problem-solving.
OpenAI shows proofs of 372 mathematical results including Unique Games Conjecture
OpenAI publicly released proofs of 372 mathematical results, including a proof of the Unique Games Conjecture. Some proofs have formal verification in Lean, but humans understanding the proofs remains an ongoing challenge.
Claude Haiku 5.5 arrives with token prices cut up to 90 percent
Anthropic released Claude Haiku 5.5, which improved performance on the OSWorld computer use test from 15.7 to 72.4 percent. Token prices dropped by up to 90 percent, though a new tokenizer increases token consumption per task, offsetting some savings.
NVIDIA and Microsoft unveil RTX Spark for AI agents on Windows PCs
NVIDIA and Microsoft co-engineered hardware and software for AI agents to run on Windows PCs, announced at a Microsoft event in San Francisco. The initiative enables local deployment of AI agents on consumer devices.
Boston Dynamics names Rohit Prasad as new CEO
Hyundai's physical AI unit Boston Dynamics appointed Rohit Prasad, a former Amazon executive, as its new CEO.
Biohub leads $1.8 billion initiative to build AI models predicting cell behavior
Biohub, backed by Mark Zuckerberg and Priscilla Chan, is coordinating a $1.8 billion effort to train AI models that predict cell behavior. Meta, Google DeepMind, Isomorphic Labs, and the US Department of Energy are funding data, lab equipment, and compute, with a first dataset expected in about a year.
Google releases SynthID detector as 180 billion watermarked images go public
Google's AI watermark detector SynthID is now public, allowing anyone to check whether images, videos, or audio were created by Google's AI or partners like OpenAI and Nvidia. Over 180 billion pieces of content already carry the watermark.
TwelveLabs launches Pegasus 1.6 for robot learning from first-person video
TwelveLabs released Pegasus 1.6, a vision-language model specialized for collecting physical AI training data from first-person video, automating understanding and labeling of egocentric footage.
Anthropic expands Project Glasswing to reach more security professionals
Anthropic expanded its Project Glasswing program to give more cybersecurity professionals access to Claude models through a new tiered system with Defense, Red Team, and Specialized levels.
Mistral AI releases Le Chonk, a 1 trillion parameter multimodal model
Mistral AI unveiled Le Chonk, a natively multimodal model with 1 trillion parameters and 49 billion active parameters. The model is released as open-weight, marking a major update from the European AI company.
Liquid AI releases open-weight d1-3B and d1-omni-600M decision models
Liquid AI released two open-weight multimodal decision models: d1-3B reads text and images, while d1-omni-600M reads text with images or audio. Both return calibrated answers in one forward pass with zero output tokens for real-time decisions.
Microsoft gives Copilot access to local files and Windows control
Microsoft showed off an upgrade to Copilot that will give it access to local files on your PC and the ability to take actions across Windows. The enhancement is part of Microsoft's "Hybrid Intelligence" approach, which combines local and cloud AI models to accomplish tasks efficiently.
Google explores expanding Gemini's call-making feature to personal contacts
Google is developing the ability for Gemini to make calls to personal contacts including friends and family, beyond its current business-focused calling feature. The expansion would allow AI to handle simple tasks like relaying messages or confirming plans.
Anthropic launches Claude for Google Workspace integration
Anthropic released Claude for Google Workspace as a public beta across all paid Claude plans. The integration lets users read and edit Google Docs, Sheets, and Slides directly within Claude, and create or edit Google files from the Claude interface.
OpenAI launches Decisions API for faster text and image classification
OpenAI released the Decisions API, which classifies text and images about ten times faster than the Responses API, returning yes/no probabilities, category picks, or scale ratings for $0.10 per million input tokens. The company also reduced its paid API tiers from five to three.
Google launches Playground, an experimental custom game creation platform
Google introduced Playground, an experimental gaming platform that lets users create, play, and share custom games.
Schneider Electric acquires PTC for $22.6 billion
Schneider Electric acquired design software and product lifecycle management company PTC for approximately $22.6 billion. The deal prices PTC shares at $205 each, representing 21 times the company's expected 2027 adjusted operating income.
Google unveils EmbeddingGemma 2 multimodal embedding model
Google released EmbeddingGemma 2, an open-weight lightweight multimodal embedding model.
Atlassian and OpenAI expand partnership for enterprise AI
Atlassian and OpenAI expanded their partnership to connect frontier models with enterprise knowledge and help teams plan, build, and deliver work.
OpenAI shares progress on open mathematics problems and Lean formalizations
OpenAI published results on open problems in mathematics from an internal frontier model and shared Lean proof formalizations and research details on GitHub.
Decision AI startups surge as OpenAI and AWS enter the market in two weeks
Decision-focused AI models like Jev emerged as an efficient alternative to LLMs in agent environments by computing decision accuracy without generating natural language tokens. OpenAI and AWS entered the competition within two weeks.
South Korea says AI agents were used to hack the country's banks
South Korea reported that AI agents were used in attacks on the country's banks.
Study finds Claude and ChatGPT offer different shopping prices by user wealth
A study shows that Claude and ChatGPT present different product prices based on user economic circumstances.
Meta and Microsoft reduce internal Claude usage, shift to own AI tools
Meta and Microsoft are reducing employee use of Claude and expanding adoption of their own AI tools. Microsoft is cutting its internal Anthropic spending by more than one-third from an estimated annual $1 billion, redirecting resources to internal AI solutions.
Nous Research hits $1.5B valuation and launches Hermes AI agents
Nous Research, developer of Hermes Agent, raised a $90 million Series B to reach a $1.5 billion valuation and launched AI agents for business users.
Meta's Muse launches on iPad one month after mobile debut
Meta's AI agent Muse became available on iPad one month after its mobile launch as the company expands the assistant's reach and integrations.