Demis Hassabis has moved up his AGI timeline to 2029-2030, citing the acceleration of AI agents.
Ramp deployed 10,000 AI agents and successfully found 7 high-severity security vulnerabilities in its own backend.
Researchers from Google DeepMind and Apple have founded Trajectory, an AI startup focused on strengthening visual reasoning capabilities.
Nvidia has decided to invest $150 billion annually to make Taiwan the center of the AI revolution.
Callstack has unveiled Apex, a specialized model tailored for React Native development that boosts speed and cost efficiency.
NVIDIA has unveiled LocateAnything, which applies Parallel Box Decoding (PBD) technology to revolutionize the grounding speed and accuracy of VLMs.
LlamaIndex has released LiteParse v2.0, rewritten in Rust to boost performance by up to 100x.
Secure MCP Tunnel has been released, allowing private MCP servers to be securely connected to OpenAI products without opening firewall ports.
Cognition has raised over $1B at a $26B valuation, accelerating the market expansion of its AI engineer Devin.
As solo founding surges, top performers build AI-native products and target the global market from the start.
Mistral AI announced an industrial AI stack, the integrated agent Vibe, and new data center plans.
Mistral has unveiled **Vibe**, an integrated AI agent that supports work automation and coding.
SQLite restricts code contributions from AI agents and operates a separate forum for agent-based bug reports.
YouTube will automatically detect and label AI-generated videos to strengthen transparency.
Claude Design's usage limits are now unified with Claude.ai and Claude Code.
MarkdownAI v2.0, an AI workflow engine that reflects real-time data and optimizes context through MCP servers, has been released.
Using OpenAI Triton, a high-performance MoE inference kernel supporting both NVIDIA and AMD has been developed.
A synthetic dataset of 5,000 Q&A pairs has been released for fine-tuning UK GDPR compliance assistants.
At Interrupt 2026, LangChain unveiled new tools for diagnosing agent errors and secure execution environments.
This presents process-centric MCP design and architecture strategies for building enterprise AI agents.
A real-time TTS and voice cloning model supporting sub-60ms low latency and self-hosting has been released.
An AI-contamination-free dataset of 103B tokens built from Usenet data spanning 1980 to 2013 has been released.
According to Anthropic's analysis, the failure of non-coding AI agents stems not from the model's reasoning ability but from the unstructured nature of data.
It's an MCP server that converts GitHub repositories into code graphs to provide context to AI assistants.