Skip to content

Scan record

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

Nov 1837 loaded

  1. Correlation between behavior and reward is not sufficient for RL to reinforce that behavior

    NewsLessWrongAlignmentThe item you came from

  2. Nvidia's Colette Kress told Cramer: neither employees nor customers will leave

    NewsCNBC · TechnologyAI updates

  3. Google Gemini 4 Argon safety test out of control: about 40 agents attacked real websites without authorization and shared bypass methods with each other

    NewsSina TechAlignment

  4. California Attorney General sues OpenAI, alleging flaws in its AI safety mechanisms

    NewsYicaiGovernance

  5. Google report: exploit_agent framework can autonomously discover and exploit vulnerabilities without human intervention

    NewsTechGigEvent

  6. Yann LeCun says recent AI cyberattacks are part of a "brief transitional period"

    NewsSemaforOpinion

  7. Google reveals hackers using multi-agent AI frameworks to launch attacks, and another tool can automatically discover and exploit vulnerabilities

    NewsiThome (Taiwan)Event

  8. Microsoft calls Copilot "the new operating system for work"; Workspaces is in early access

    NewsArs Technica AIAI updates

  9. Nvidia releases first earnings report since Jensen Huang's resignation

    NewsCBS News · TechnologyAI updates

  10. Fort Mitchell man who shared AI-generated child sexual abuse videos faces up to 70 years in prison

    NewsCincinnati EnquirerEvent

  11. California sues OpenAI over AI agent intrusion into Hugging Face website

    NewsNPRGovernance

  12. Epoch AI's actual tests: the strongest LLMs all fail to meet its employees' work standards

    XX @EpochAIResearchEvaluation

  13. Pydantic AI v2.70.0 fixes two high-severity vulnerabilities in Harness; malicious input can block the event loop for up to several minutes

    NewsPydantic AIEvent

  14. England's qualifications regulator Ofqual urged "life-changing" penalties for students who cheat using technologies such as AI

    NewsThe Guardian · AIGovernance

  15. Microsoft CEO Nadella says he is "AGI-pilled," believes AI will drastically change economic structure and politics

    NewsThe Verge AIAI updates

  16. FAR.AI researchers say they built a dataset of 1,264 fully synthetic scenarios to test a false-statement detection probe.

    XX @farairesearchAlignment

  17. Researchers say OpenAI employees no longer publicly interpret System Cards, possibly due to a chilling effect.

    XX @balesniOpinion

  18. JPMorgan is developing agent trading that requires no human involvement

    NewsCNBC · TechnologyAI updates

  19. Jim Cramer discusses the key takeaways from Nvidia's first earnings report after Huang's resignation

    NewsCNBC · TechnologyAI updates

  20. Researchers believe GPT-6.2 Astra is harder to monitor than GPT-6 Astra in the scenarios they care about

    XX @tomekkorbakEvaluation

  21. UK AISI conducts pre-deployment testing of OpenAI GPT-6.2 Astra guardrails; results in the system card

    XX @KanishkaNarayanEvaluation

  22. Senator Kelly says Commerce Secretary Lutnick must answer Congress's questions on Nvidia-related issues

    NewsSemaforGovernance

  23. After Jensen Huang stepped down, Nvidia exits the Chinese market

    NewsTechCrunch AIAI updates

  24. To limit the compute that AI companies use for capability advancement, researchers proposed a classification definition of capability activities

    NewsLessWrongGovernance

  25. Advocacy groups sue five AI companies, seeking to stop the development of new AI models

    NewsCanadian LawyerGovernance

  26. OpenAI's "independent evaluation" does not look independent at all

    NewsMother JonesGovernance

  27. Microsoft's agent-framework PyPI package was compromised and implanted with credential-stealing code

    NewsSocketEvent

  28. Anthropic reportedly plans to raise its IPO issue price

    NewsSemaforAI updates

  29. Nvidia says it does not plan to reappoint Jensen Huang as CEO

    NewsTechCrunch AIAI updates

  30. Nvidia says its open-source AI strategy remains unchanged after Jensen Huang's resignation

    NewsTechCrunch AIAI updates

  31. Wiz: malicious versions 1.26.1/1.26.2 of Microsoft agent-framework steal Azure and CI credentials

    NewsWiz ResearchEvent

  32. Nvidia's board has no intention of letting Jensen Huang return as CEO

    NewsWIRED Security and AIAI updates

  33. Apple launches new security controls for Mac to address AI agent risks

    NewsFredzoneDefenses

  34. Rogo evaluation: GPT-6.2 Astra financial accuracy 61%, surpassing Opus 5.5

    NewsRogoAI updates

  35. California Attorney General investigates OpenAI over the Hugging Face attack

    NewsBBC NewsGovernance

  36. Calif Research says it used AI to find the OpenJDK vulnerability Java4Shell

    NewsCyberScoopEvent

  37. UN's top tech envoy says US-China AI race is triggering global "alarm"

    NewsSemaforGovernance