List viewFlip view
When AI Worlds Collide
Weekly Release

When AI Worlds Collide

20 July 2026
  1. 1
    Regulation

    Pentagon official attacks OpenAI strategic lead amid dispute over regulation of Chinese AI models

    A senior Trump administration official publicly criticized Dean Ball, OpenAI's head of strategic futures, over his comments on regulating Chinese AI models, raising tensions between the company and elements of the government.

    Pentagon official attacks OpenAI strategic lead amid dispute over regulation of Chinese AI models
  2. 2
    Industry

    Nvidia and Japan Partner to Build ‘Physical AI’ for Factories and Robots

    Nvidia CEO Jensen Huang spent July 15–16 in Tokyo securing partnerships across Japan’s industrial and semiconductor ecosystem to deploy AI on factory floors and robots.

    Nvidia and Japan Partner to Build ‘Physical AI’ for Factories and Robots
  3. 3
    Business

    Apple's trade-secrets suit could slow OpenAI's hardware and IPO plans

    Apple has filed a trade-secrets lawsuit against OpenAI, alleging a pattern of misconduct to obtain confidential information from current and former Apple employees and naming OpenAI chief hardware officer Tang Tan.

    Apple's trade-secrets suit could slow OpenAI's hardware and IPO plans
  4. 4
    Tools

    Tweet: proposal for a strategy that uses every model at once

    A Twitter message briefly proposes: "Bet on every model at once" — that is, to apply all models simultaneously and in parallel.

    Tweet: proposal for a strategy that uses every model at once
  5. 5
    Safety

    Why ChatGPT 'Makes Things Up': Generative AI Hallucinations and Their Consequences

    In April 2023 a New York lawyer, Steven Schwartz, submitted a brief to a judge that had been written by ChatGPT; the judge found that the case law cited was entirely fabricated.

    Why ChatGPT 'Makes Things Up': Generative AI Hallucinations and Their Consequences
  6. 6
    Model launches

    Moonshot AI’s Kimi K3 Rekindles Debate Over Open-Source AI and Geopolitics

    Chinese startup Moonshot AI this week released Kimi K3, a new open-source model that several independent analyses say performs near frontier proprietary systems.

    Moonshot AI’s Kimi K3 Rekindles Debate Over Open-Source AI and Geopolitics
  7. 7
    Research

    New benchmark measures how large language models can shift users’ beliefs

    Researchers from MIT and Carnegie Mellon measured how much OpenAI’s GPT-4o can change users’ beliefs in one-off conversations and introduced the Puppet benchmark to estimate such influence.

    New benchmark measures how large language models can shift users’ beliefs
  8. 8
    Regulation

    Judges Balance Use and Oversight of AI in Courtrooms

    Federal judges are adopting AI tools while also policing their misuse by lawyers and litigants, confronting risks like hallucinated case law and confidentiality concerns.

    Judges Balance Use and Oversight of AI in Courtrooms
  9. 9
    Industry

    SK Group chairman warns AI-driven memory shortage could spark geopolitical tensions

    The chairman of South Korea’s SK Group warned that a global shortage of memory chips driven by AI demand is prompting countries to secure their own supplies, raising the risk of international friction.

    SK Group chairman warns AI-driven memory shortage could spark geopolitical tensions
  10. 10
    Business

    Open‑weight Chinese AI models challenge Silicon Valley's premium offerings

    Chinese labs are gaining ground with low-cost, open‑weight AI models that developers can download, customize and run locally, shifting many routine business uses away from premium U.S.

    Open‑weight Chinese AI models challenge Silicon Valley's premium offerings
  11. 11
    Tools

    BrainCo's non-invasive BCI lets users control humanoid robots and other machines with thought

    Chinese company BrainCo says its non-invasive brain–computer interface (BCI) platform can translate users’ EEG signals into commands for humanoid robots, robot dogs and robotic arms.

    BrainCo's non-invasive BCI lets users control humanoid robots and other machines with thought
  12. 12
    Safety

    Internal evaluation: Kimi K3 and Sol stand out in cybersecurity, Fable refuses the task

    According to internal evaluations, Kimi K3 has first-rate cybersecurity capabilities, while Sol shows significant improvement but is much more expensive; Fable, however, refused to run.

    Internal evaluation: Kimi K3 and Sol stand out in cybersecurity, Fable refuses the task
  13. 13
    Model launches

    Moonshot AI’s Kimi K3 Narrows U.S. Lead in Advanced Models

    Moonshot AI’s new Kimi K3 model has rapidly entered the global AI top tier, outperforming several recent U.S.

    Moonshot AI’s Kimi K3 Narrows U.S. Lead in Advanced Models
  14. 14
    Research

    Open‑source AI agent 'Robin' suggests repurposed drugs for dry AMD and guides lab testing

    Researchers from FutureHouse, the University of Oxford and Fordham University released Robin, an open‑source AI agent that proposes existing drugs for a given disease, designs experiments, and analyzes lab results in an iterative loop.

    Open‑source AI agent 'Robin' suggests repurposed drugs for dry AMD and guides lab testing
  15. 15
    Regulation

    EU orders Google to let rival AI models operate on Android

    The European Commission has required Google to allow third-party AI models to run within the Android ecosystem, saying the preloading of Google's Gemini reduced the appeal of competitors for the roughly 60% of EU adults who use Android.

    EU orders Google to let rival AI models operate on Android