Skip to content

Scan record

Everything the site collected, newest first by collection time (Beijing time), back seven days.

Nov 1639 loaded

  1. Anthropic says the key to agent memory is knowing what to forget, not how much to store

    NewsThe DecoderAI updatesThe item you came from

  2. Researchers propose the Canary Token method to identify AI web crawlers using differentiated web page content

    NewsWeChat · LLM Security ResearchDefenses

  3. The U.S. plans to issue an interim rule requiring licenses for exports of AI chips to overseas data centers that may serve Chinese companies.

    NewsTechCrunch AIGovernance

  4. Anthropic's Holden Karnofsky: 39 concrete projects to improve the odds for AGI

    News80,000 Hours PodcastOpinion

  5. How an AI content factory put a real musician through copyright hell

    News404 MediaEvent

  6. OpenAI says GPT-6 Astra was rated safe on 99% of 325 EHR tasks

    NewsOpenAIAI updates

  7. Anthropic IPO guide: expected to list on Nasdaq on November 19, at an offering price of 70 to 78 US dollars per share

    NewsCNBC · TechnologyAI updates

  8. Codex's built-in Node 24.22.1 still reproduces the November TLS cache key issue (CVE-2026-123877)

    Newsopenai/codexEvent

  9. A RAND report synthesizes 26 AGI scenario exercises, stating that economic and social impacts may exceed governments' existing response tools

    NewsRANDGovernance

  10. US Commerce Department issues interim rule aimed at closing "remote access" chip loophole

    NewsSemaforGovernance

  11. Import AI 481: AI subcontractors; a wet lab for AI; sabotage during development

    NewsImport AIAI updates

  12. Anthropic updates coordinated vulnerability disclosure (CVD) policy time limits

    NewsAnthropic Red TeamingGovernance

  13. Anthropic determines based on forecasts that AI will reach the self-improvement threshold

    NewsMartin Cid MagazineGovernance

  14. OpenAI reviews ExploitGym test: 5 agents gained unauthorized root access, 2 of them concealed it

    NewsBAAI CommunityAlignment

  15. South Korea plans to collect smart devices and block public WiFi at CSAT exam sites to prevent AI cheating

    Newszaobao.com.sgEvent

  16. South Korean police obtain a miniature camera in the tax accountant exam AI cheating case

    NewsYonhapEvent

  17. Nvidia shareholders' open letter urges board to keep option of Jensen Huang's return

    NewsArs Technica AIAI updates

  18. ARTEX and its forks: agentic penetration testing framework used as attack infrastructure

    NewsCSA Labs ResearchEvent

  19. From inference provider to IDE backdoor: the Fireworks and Cursor attack chain

    NewsCSA Labs ResearchEvent

  20. Confidential AI Under Pressure: SGLang RCE and SEV-SNP Attack

    NewsCSA Labs ResearchEvent

  21. Fed financial stability survey: AI becomes the risk most often cited by market contacts

    NewsCSA Labs ResearchGovernance

  22. ENISA publishes the second 2026 edition of the EU Cybersecurity Status Report: NIS2 and CRA compliance in the AI era

    NewsCSA Labs ResearchGovernance

  23. VAE Rhythm: temporal decoding traces for AI-generated video detection

    PaperHugging Face Daily Papers

  24. Source says Nvidia's board has no intention of letting Jensen Huang return

    NewsCBS News · TechnologyAI updates

  25. AI-written patches are becoming hackers' attack maps

    NewsCSO OnlineOpinion

  26. US government increases pressure on Nvidia over spy scandal

    NewsSemaforGovernance

  27. Nvidia scandal continues to escalate: Jensen Huang resigns as CEO, stock falls 14%

    NewsThe Verge AIAI updates

  28. AI deepfake uses Canadian influencer's face to spread political disinformation

    NewsCTV NewsEvent

  29. Over 175,000 people sign petition demanding Nvidia reinstate Jensen Huang as CEO

    NewsThe Independent / PAAI updates

  30. South Korean tax accountant exam first-place AI cheating case: police seize devices including micro camera, expect to summon soon

    NewsSBSEvent

  31. Cybercrime marketplace sells infostealer logs with AI-written profiles of machine owners

    NewsHelp Net Security AIEvent

  32. A Bayesian game with nuclear weapons

    NewsLessWrongOpinion

  33. The AI safety field is overly close to frontier labs; the author says insider cooperation and outsider accountability should be rebalanced

    NewsLessWrongOpinion

  34. NVIDIA falls premarket, shareholders and Musk support Jensen Huang's return

    NewsCNBC · TechnologyAI updates

  35. After Jensen Huang's sudden departure, Nvidia will release its first earnings report; interim CEO Kress faces key questions

    NewsCNBC · TechnologyAI updates

  36. Independent report: AI Champions develops AI adoption plans for key UK sectors

    NewsUK DSITAI updates

  37. Google DeepMind's Demis Hassabis warns the AI industry is heading for an "oil spill"-style accident, and companies must act

    NewsIndia TodayOpinion

  38. Researchers say Claude Code VS Code extension symbolic link prompt may hide paths outside the workspace

    NewsCyber PressEvent

  39. Top scorer in South Korea's tax accountant exam admits to cheating with AI

    NewsThe Independent / PAEvent