| raw_data |
{"id": 509526, "content": "【An {"id": 509526, "content": "【Anthropic Claude Mythos在Firefox浏览器发现271个安全漏洞,相关漏洞已于本周完成修复】金色财经报道,4月22日,Mozilla近日披露,Anthropic旗下最新AI模型Claude Mythos在对Firefox浏览器的内部测试中识别出271个安全漏洞,相关漏洞已于本周完成修复。 \n作为对比,此前另一版本Anthropic模型仅发现22个安全敏感漏洞。Mozilla表示,所有被发现的漏洞均未超出顶尖人类研究员的发现范畴。Claude Mythos于2026年3月正式发布,是Anthropic在推理、编码及网络安全领域最强大的模型,目前通过“Project Glasswing”计划限定向亚马逊、苹果、微软等经审核的合作伙伴开放使用。", "content_prefix": "Anthropic Claude Mythos在Firefox浏览器发现271个安全漏洞,相关漏洞已于本周完成修复", "link_name": "原文链接", "link": "https://decrypt.co/365099/anthropic-claude-mythos-ai-271-vulnerabilities-firefox-seriously-powerful", "grade": 4, "sort": "", "category": 0, "highlight_color": "", "images": [], "created_at": 1776824280, "created_at_zh": "2026-04-22", "attribute": "", "up_counts": 350, "down_counts": 306, "zan_status": "", "readings": [], "extra_type": 0, "extra": null, "prev": null, "next": null, "word_blocks": [], "is_show_comment": 1, "is_forbid_comment": 0, "comment_count": 0, "analyst_user": null, "show_source_name": "", "vote_id": 0, "vote": null, "date": "2026-04-22"}... |