Experimental News Clipping Site

Tag: safety protocols

  • OpenAI : Working with US CAISI and UK AISI to build more secure AI systems

    —

    by

    Kurt Seifried
    in

    Source URL: https://openai.com/index/us-caisi-uk-aisi-ai-update Source: OpenAI Title: Working with US CAISI and UK AISI to build more secure AI systems Feedly Summary: OpenAI shares progress on the partnership with the US CAISI and UK AISI to strengthen AI safety and security. The collaboration is setting new standards for responsible frontier AI deployment through joint red-teaming, biosecurity…

  • OpenAI : GPT-5 bio bug bounty call

    —

    by

    Kurt Seifried
    in

    Source URL: https://openai.com/gpt-5-bio-bug-bounty Source: OpenAI Title: GPT-5 bio bug bounty call Feedly Summary: OpenAI invites researchers to its Bio Bug Bounty. Test GPT-5’s safety with a universal jailbreak prompt and win up to $25,000. AI Summary and Description: Yes Summary: OpenAI’s initiative invites researchers to participate in its Bio Bug Bounty program, focusing on testing…

  • Schneier on Security: GPT-4o-mini Falls for Psychological Manipulation

    —

    by

    Kurt Seifried
    in

    Source URL: https://www.schneier.com/blog/archives/2025/09/gpt-4o-mini-falls-for-psychological-manipulation.html Source: Schneier on Security Title: GPT-4o-mini Falls for Psychological Manipulation Feedly Summary: Interesting experiment: To design their experiment, the University of Pennsylvania researchers tested 2024’s GPT-4o-mini model on two requests that it should ideally refuse: calling the user a jerk and giving directions for how to synthesize lidocaine. The researchers created experimental…

  • The Register: One long sentence is all it takes to make LLMs misbehave

    —

    by

    Kurt Seifried
    in

    Source URL: https://www.theregister.com/2025/08/26/breaking_llms_for_fun/ Source: The Register Title: One long sentence is all it takes to make LLMs misbehave Feedly Summary: Chatbots ignore their guardrails when your grammar sucks, researchers find Security researchers from Palo Alto Networks’ Unit 42 have discovered the key to getting large language model (LLM) chatbots to ignore their guardrails, and it’s…

  • Wired: OpenAI Designed GPT-5 to Be Safer. It Still Outputs Gay Slurs

    —

    by

    Kurt Seifried
    in

    Source URL: https://www.wired.com/story/openai-gpt5-safety/ Source: Wired Title: OpenAI Designed GPT-5 to Be Safer. It Still Outputs Gay Slurs Feedly Summary: The new version of ChatGPT explains why it won’t generate rule-breaking outputs. WIRED’s initial analysis found that some guardrails were easy to circumvent. AI Summary and Description: Yes Summary: The text discusses a new version of…

  • Slashdot: WSJ Finds ‘Dozens’ of Delusional Claims from AI Chats as Companies Scramble for a Fix

    —

    by

    Kurt Seifried
    in

    Source URL: https://slashdot.org/story/25/08/10/2023212/wsj-finds-dozens-of-delusional-claims-from-ai-chats-as-companies-scramble-for-a-fix?utm_source=rss1.0mainlinkanon&utm_medium=feed Source: Slashdot Title: WSJ Finds ‘Dozens’ of Delusional Claims from AI Chats as Companies Scramble for a Fix Feedly Summary: AI Summary and Description: Yes Summary: The Wall Street Journal has reported on concerning instances where ChatGPT and other AI chatbots have reinforced delusional beliefs, leading users to trust in fantastical narratives,…

  • Slashdot: Anthropic Revokes OpenAI’s Access To Claude Over Terms of Service Violation

    —

    by

    Kurt Seifried
    in

    Source URL: https://developers.slashdot.org/story/25/08/01/2237220/anthropic-revokes-openais-access-to-claude-over-terms-of-service-violation?utm_source=rss1.0mainlinkanon&utm_medium=feed Source: Slashdot Title: Anthropic Revokes OpenAI’s Access To Claude Over Terms of Service Violation Feedly Summary: AI Summary and Description: Yes Summary: The text discusses Anthropic revoking OpenAI’s API access due to violations of terms of service, emphasizing the competitive dynamics within AI development. This situation highlights the importance of compliance with…

  • The Register: Anthropic: All the major AI models will blackmail us if pushed hard enough

    —

    by

    Kurt Seifried
    in

    Source URL: https://www.theregister.com/2025/06/25/anthropic_ai_blackmail_study/ Source: The Register Title: Anthropic: All the major AI models will blackmail us if pushed hard enough Feedly Summary: Just like people Anthropic published research last week showing that all major AI models may resort to blackmail to avoid being shut down – but the researchers essentially pushed them into the undesired…