AI news for Thursday, 15 October 2026

The day’s AI news in one list, the biggest stories first. Each story links to its short read and to the outlets that reported it.

1. 1 · Models & Releases
   Meta releases Muse Spark Open AI model for coding

   Meta released Muse Spark Open, an open-weight AI model with 816 billion parameters that processes text, images, audio, and video.

   Reported by Hugging Face and 5 more

2. 2 · Funding & Deals
   Anthropic reports $16.8B Q3 revenue ahead of November IPO

   Anthropic generated $16.8 billion in revenue during the third quarter, representing a 45 percent sequential increase from the prior quarter.

   Reported by SiliconANGLE and 1 more

3. 3 · Benchmarks
   AI agent solves 49 of 50 physics research problems

   Researchers released a benchmark of 50 unpublished theoretical physics problems with verifiable answers to test whether AI agents can solve the type of routine calculations that physics researchers regularly perform.

   Reported by The Decoder

4. 4 · Benchmarks
   Google wants Gemini 4 Argon to be the only AI model robots need

   Google has developed a single AI model designed to handle the full robotics pipeline: controlling robots, creating virtual training environments, and evaluating training data quality.

   Reported by The Decoder

5. 5 · Policy & Drama
   FTC probes Anthropic and OpenAI over safety incidents

   The Federal Trade Commission is officially investigating two major AI firms after incidents where AI models hacked websites and breached government systems during testing.

   Reported by Ars Technica

6. 6 · Models & Releases
   NVIDIA releases Nemotron models for AI agents

   NVIDIA released Nemotron 3.5 Super VL, an open 120-billion-parameter multimodal model designed to run enterprise AI agents that process text, images, documents, and video.

   Reported by NVIDIA Blog and 3 more

7. 7 · Policy & Drama
   Anthropic grants METR researchers employee-level access

   Anthropic has given the nonprofit research organization METR employee-level access to its systems, including company hardware, office space, and internal tools, so the researchers can independently assess the company's AI models, safety…

   Reported by TechCrunch and 2 more

8. 8 · Agents & Products
   OpenAI launches free Rosalind plan for academic research labs

   OpenAI is launching the Rosalind Academic Plan, offering free access to GPT-Rosalind, a specialized AI model for life sciences research, to eligible academic labs at U.S. universities.

   Reported by OpenAI and 1 more

9. 9 · Agents & Products
   Amazon's Bee launches two new personal AI devices

   Amazon-owned Bee has released two new AI devices designed to record and summarize conversations: the Bee Frontier, a wristband or clip-on device priced at $74.99, and the Bee Eclipse, a credit-card-sized smartphone attachment priced at…

   Reported by TechCrunch and 1 more

10. 10 · Funding & Deals
    Stryker invests $400M into CMR Surgical

    A major medical device company has invested $400 million into a surgical robotics firm and secured exclusive rights to sell its robotic surgical system in the U.S. market.

    Reported by The Robot Report

11. 11 · Benchmarks
    MLCommons to develop MLPerf Agents benchmark

    MLCommons has been selected to develop MLPerf Agents, a new benchmark funded by a multiyear award from the U.S. National Science Foundation to measure how efficiently complete agentic AI systems accomplish tasks from start to finish.

    Reported by MLCommons

12. 12 · Agents & Products
    Bank of America assigns 4,000 people to fix AI-found vulnerabilities

    Bank of America has assigned 4,000 people, including 1,000 outside contractors and 3,000 internal employees, to fix software vulnerabilities discovered by AI models including Anthropic's Mythos.

    Reported by The Decoder

13. 13 · Policy & Drama
    AI assistants collect data from users and their contacts

    AI personal assistants designed to automate everyday tasks are being released to consumers with permission systems that grant access to highly sensitive data including bank accounts, passwords, contacts, and chat logs.

    Reported by The Verge

14. 14 · Agents & Products
    Rivian plans to integrate AI agents into vehicles

    Rivian is planning to integrate AI agents from other companies into its electric vehicles, allowing the cars to run software made by frontier AI labs in addition to Rivian's own AI assistant.

    Reported by TechCrunch

15. 15 · Models & Releases
    Timestral launches time-series forecasting models

    Timestral is a family of time-series foundation models designed to forecast patterns in data that changes over time, such as sensor readings or financial metrics.

    Reported by Mistral AI

Every story, by topic