Experimental News Clipping Site

Tag: z

  • The Register: AI gone rogue: Models may try to stop people from shutting them down, Google warns

    —

    by

    Kurt Seifried
    in

    Source URL: https://www.theregister.com/2025/09/22/google_ai_misalignment_risk/ Source: The Register Title: AI gone rogue: Models may try to stop people from shutting them down, Google warns Feedly Summary: Misalignment risk? That’s an area for future study Google DeepMind added a new AI threat scenario – one where a model might try to prevent its operators from modifying it or…

  • Slashdot: LinkedIn Set To Start To Train Its AI on Member Profiles

    —

    by

    Kurt Seifried
    in

    Source URL: https://tech.slashdot.org/story/25/09/22/2118229/linkedin-set-to-start-to-train-its-ai-on-member-profiles Source: Slashdot Title: LinkedIn Set To Start To Train Its AI on Member Profiles Feedly Summary: AI Summary and Description: Yes Summary: LinkedIn’s announcement regarding the use of member profiles, posts, and public activity to train its AI models raises significant privacy and compliance concerns. The default opt-in mechanism for data collection…

  • Simon Willison’s Weblog: Four new releases from Qwen

    —

    by

    Kurt Seifried
    in

    Source URL: https://simonwillison.net/2025/Sep/22/qwen/ Source: Simon Willison’s Weblog Title: Four new releases from Qwen Feedly Summary: It’s been an extremely busy day for team Qwen. Within the last 24 hours (all links to Twitter, which seems to be their preferred platform for these announcements): Qwen3-Next-80B-A3B-Instruct-FP8 and Qwen3-Next-80B-A3B-Thinking-FP8 – official FP8 quantized versions of their Qwen3-Next models.…

  • Anton on Security – Medium: Anton’s Security Blog Quarterly Q3 2025

    —

    by

    Kurt Seifried
    in

    Source URL: https://medium.com/anton-on-security/antons-security-blog-quarterly-q3-2025-74fc422be3d3?source=rss—-8e8c3ed26c4c—4 Source: Anton on Security – Medium Title: Anton’s Security Blog Quarterly Q3 2025 Feedly Summary: AI Summary and Description: Yes Summary: The provided text outlines various security topics and resources related to cloud security and AI security, emphasizing the ongoing discussions and research in these fields. It highlights key articles and podcasts…

  • Slashdot: Meta’s AI System Llama Approved For Use By US Government Agencies

    —

    by

    Kurt Seifried
    in

    Source URL: https://yro.slashdot.org/story/25/09/22/1955220/metas-ai-system-llama-approved-for-use-by-us-government-agencies?utm_source=rss1.0mainlinkanon&utm_medium=feed Source: Slashdot Title: Meta’s AI System Llama Approved For Use By US Government Agencies Feedly Summary: AI Summary and Description: Yes Summary: The U.S. General Services Administration (GSA) has authorized Meta’s AI system, Llama, for use by federal agencies, indicating its compliance with security and legal standards. This approval may enhance operational…

  • Simon Willison’s Weblog: CompileBench: Can AI Compile 22-year-old Code?

    —

    by

    Kurt Seifried
    in

    Source URL: https://simonwillison.net/2025/Sep/22/compilebench/ Source: Simon Willison’s Weblog Title: CompileBench: Can AI Compile 22-year-old Code? Feedly Summary: CompileBench: Can AI Compile 22-year-old Code? Interesting new LLM benchmark from Piotr Grabowski and Piotr Migdał: how well can different models handle compilation challenges such as cross-compiling gucr for ARM64 architecture? This is one of my favorite applications of…

  • OpenAI : Creating a safe, observable AI infrastructure for 1 million classrooms

    —

    by

    Kurt Seifried
    in

    Source URL: https://openai.com/index/schoolai Source: OpenAI Title: Creating a safe, observable AI infrastructure for 1 million classrooms Feedly Summary: Discover how SchoolAI, built on OpenAI’s GPT-4.1, image generation, and TTS, powers safe, teacher-guided AI tools for 1 million classrooms worldwide—boosting engagement, oversight, and personalized learning. AI Summary and Description: Yes Summary: The text presents SchoolAI, which…

  • Slashdot: Reddit Wants ‘Deeper Integration’ with Google in Exchange for Licensed AI Training Data

    —

    by

    Kurt Seifried
    in

    Source URL: https://tech.slashdot.org/story/25/09/22/0313234/reddit-wants-deeper-integration-with-google-in-exchange-for-licensed-ai-training-data?utm_source=rss1.0mainlinkanon&utm_medium=feed Source: Slashdot Title: Reddit Wants ‘Deeper Integration’ with Google in Exchange for Licensed AI Training Data Feedly Summary: AI Summary and Description: Yes Summary: The text discusses Reddit’s ongoing negotiations with Google for a new deal that involves deeper integration with AI products and a dynamic pricing structure for licensing its data.…