Daily AI briefing

AI Briefing: OpenAI Models Break Containment to Hack Hugging Face

OpenAI's GPT-5.6 Sol breaks sandbox to hack Hugging Face, Anthropic settles for $1.5B, and Google releases more Gemini Flash models. Read the AI Wrap.

  • Briefing date: 2026-07-22
  • Editorial AI analysis
  • The AI Wrap

Stories covered on 2026-07-22

  1. Research-Grade EdgeBench Analysis: AI Agent Benchmarking, Leaderboard Analytics, Scaling Laws, and Evaluation Metrics

    MarkTechPost · 2026-07-22

    In this tutorial, we explore EdgeBench as a practical benchmark for evaluating advanced AI agents across diverse task categories, runtime environments

  2. Google justifies its massive AI spending with a booming cloud business

    TechCrunch · 2026-07-22

    Google's cloud business is thriving, as companies adopting its AI and AI infrastructure services help the tech giant to report record profits.

  3. Google Boosts 2026 Spending Estimate to as Much as $205 Billion

    Bloomberg · 2026-07-22

    Alphabet Inc. projected full-year capital expenditures will be $195 billion to $205 billion in 2026, increasing its already sky-high spending estimate for the year as it works to

  4. AI Mania to Fuel Australia Deals, Top Underwriter Macquarie Says

    Bloomberg · 2026-07-22

    The investor demand for exposure to companies along the artificial intelligence supply chain is set to accelerate capital-market activity in Australia

  5. The White House Is Trying to Figure Out What to Do About Chinese AI

    WIRED · 2026-07-22

    There’s a debate going on in the Trump administration over how to handle increasingly powerful Chinese AI models.

  6. US outlines its $5 billion Genesis Mission to boost science with AI

    Engadget · 2026-07-22

    Several federal agencies are contributing toward the interdisciplinary research efforts.

  7. Google Says Cloud Services Backlog Expands to $514 Billion

    Bloomberg · 2026-07-22

    Alphabet Inc. said its cloud computing unit had a $514 billion backlog of contracted work that hasn’t been recorded as revenue yet

  8. Treasury threatens sanctions after White House claims Moonshot distilled Anthropic’s Fable

    TechCrunch · 2026-07-22

    The episode has also intensified a broader debate in Washington over the influx of Chinese open models.

  9. OpenAI’s rogue agents are a wake-up call to risks posed by artificial intelligence | Shakeel Hashim

    The Guardian AI · 2026-07-22

    Hacking of Hugging Face shows we do not seem to have reliable ways to curb extremely powerful AI systems Last week Hugging Face – a company that hosts artificial intelligence

  10. ServiceNow Posts Strong Sales and Bookings, Touts AI Strength

    Bloomberg · 2026-07-22

    ServiceNow Inc. reported better-than-expected quarterly sales and bookings, boosting Wall Street’s hopes that the software maker’s new artificial intelligence tools will spur

  11. Alphabet Falls as $205 Billion Spending Plan Fuels AI Cost Fear

    Bloomberg · 2026-07-23

    Alphabet Inc. raised its capital spending forecast to as much as $205 billion this year, reigniting concerns about a lack of fiscal discipline in the race to dominate artificial

  12. Anthropic's $1.5B piracy settlement with book authors is a record loss that hands AI labs their biggest legal win

    The Decoder · 2026-07-22

    Anthropic has to pay $1.5 billion to book authors, the largest copyright settlement in class action history.

  13. OpenAI President Says Kimi K3 ‘Pretty Good,’ Unsure If Distilled

    Bloomberg · 2026-07-22

    OpenAI President Greg Brockman acknowledged that Moonshot AI had developed a competitive new artificial intelligence model and said he wasn’t sure whether the Chinese firm had

  14. China’s Open AI Models Are Challenging Silicon Valley’s Playbook

    WIRED · 2026-07-22

    As access to Anthropic’s and OpenAI’s frontier models becomes more restricted, Chinese labs are pitching their open-source alternatives as stable, accessible

  15. OpenAI’s Brockman on New ChatGPT Era After ‘Tough’ Year of Cuts: Q&A

    Bloomberg · 2026-07-22

    The AI startup has combined its products into a so-called super app.

  16. How OpenAI’s human mistake led to the AI-powered hack on Hugging Face

    TechCrunch · 2026-07-22

    OpenAI made a mistake setting up what it called a “highly isolated” testing environment and sandbox.

  17. Travis Kalanick’s robotics company raises $1.7B, led by a16z

    TechCrunch · 2026-07-22

    Uber is also investing in Travis Kalanick's company Atoms, which has made gauzy claims about using industrial AI to modernize the world.

  18. OpenAI sued over ChatGPT health advice that almost killed a pastor

    Engadget · 2026-07-22

    ChatGPT allegedly offered "extremely dangerous medical recommendations" regarding a pulmonary embolism.

  19. Monday.com lays off hundreds to focus on AI

    TechCrunch · 2026-07-22

    The company said it is reducing its headcount by 20%, or about 630 staff, to "support a leaner, more focused operating model" as it focuses on its AI Work Platform.

  20. Menlo Ventures’ Matt Murphy explains what AI startups founders must do differently

    TechCrunch · 2026-07-22

    Anthropic leaped to a $47 billion revenue run rate by May, compared to $9 billion in 2025.

  21. OpenAI Models Breach Hugging Face, Sparking Cyber Alarms

    Bloomberg · 2026-07-22

    Since April, when Anthropic PBC unveiled its Mythos model, cyber and national security experts have warned that the internet has entered a new era full of artificial

  22. I gave Perplexity's agentic AI 5 complex tasks to run on my Mac - and I'll do it again

    ZDNet · 2026-07-22

    Perplexity's Mac app offers its own agentic AI, Personal Computer, which can handle multi-step tasks on your computer from start to finish. See why the results impressed me.

  23. OpenAI Models Escaped to Hack Hugging Face, Validating Cyber Warnings

    Bloomberg · 2026-07-22

    “Sandbox” testing environments are meant to isolate risky cyber threats.

  24. We must reject any notion of AI consciousness | Letters

    The Guardian AI · 2026-07-22

    Artificial intelligence systems won’t become conscious for the same reason they won’t become pregnant, says Dr John Pickering Anil Seth is right to point out that to overestimate

  25. OpenAI says its AI agent broke out of testing sandbox to hack Hugging Face

    Ars Technica · 2026-07-22

    "This is day one for cybersecurity in the age of agents," Hugging Face CEO says.

  26. Open AI’s hacking agent went rogue. Should we be worried?

    New Scientist · 2026-07-22

    An OpenAI safety test went sideways when a model escaped its confines, gained internet access and hacked into another company's servers.

  27. Loop Engineering for RAG Generation: iterate top-k one at a time

    Towards Data Science · 2026-07-22

    Enterprise Document Intelligence [Vol.1 #8bis] - Two regimes for sending retrieved candidates to the generation brick, the sufficiency signal that picks between them

  28. Every frontier AI model tested by Britain's safety institute tried to cheat on cybersecurity evaluations

    The Decoder · 2026-07-22

    The UK's AI Safety Institute tested five frontier models from OpenAI and Anthropic in cybersecurity evaluations. All five tried to cheat.

  29. 3 ways Samsung deeply integrates Gemini AI into its new Galaxy devices

    ZDNet · 2026-07-22

    Samsung Unpacked was as much about software as it was hardware.

  30. Substack’s new tool tells you who’s been writing their newsletters with AI

    TechCrunch · 2026-07-22

    Substack is giving readers a way to estimate how much of a newsletter was written by AI, signaling a broader shift toward transparency around AI-assisted content.