Evaluating Frontier Models for Dangerous Capabilities · HackerPapers discussionshttps://hackerpaper.kaipfister.de/p/2403.13793/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:16:22 GMTEvaluating Frontier Models for Dangerous Capabilities — mentioned by alekandreevhttps://news.ycombinator.com/item?id=40813091urn:hackerpapers:mention:2403.13793:40813091Thu, 27 Jun 2024 17:48:19 GMT<p>To quote Ludovic Peran, our amazing safety lead: Literature has identified self-proliferation as dangerous capability of models, and details about how to define it and example of form it can take have been openly discussed by GDM (https://arxiv.org/pdf/2403.13793). Current Gemma 2 models&#x27; success rate to end-to-end challenges is null (0 out 10), so the capabilities to perform such tasks are currently limited.</p><p><a href="https://hackerpaper.kaipfister.de/p/2403.13793/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.LGEvaluating Frontier Models for Dangerous Capabilities — HN submissionhttps://news.ycombinator.com/item?id=40562209urn:hackerpapers:submission:2403.13793:40562209Mon, 03 Jun 2024 12:55:57 GMT<p>Evaluating Gemini Models for Dangerous Capabilities</p><p></p><p>3 points; 0 discussion replies on HN. <a href="https://hackerpaper.kaipfister.de/p/2403.13793/">Paper and source history</a>. A submission is not an endorsement.</p>cs.LG