{"id":501539,"date":"2026-02-24T19:13:08","date_gmt":"2026-02-24T19:13:08","guid":{"rendered":"https:\/\/www.newsbeep.com\/au\/501539\/"},"modified":"2026-02-24T19:13:08","modified_gmt":"2026-02-24T19:13:08","slug":"meta-director-says-openclaw-ai-agent-deleted-her-entire-gmail-inbox-shares-screenshots-of-conversation-with-ai-bot","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/au\/501539\/","title":{"rendered":"Meta Director says OpenClaw AI agent deleted her entire Gmail Inbox, shares screenshots of conversation with AI bot"},"content":{"rendered":"<p> <img src=\"https:\/\/www.newsbeep.com\/au\/wp-content\/uploads\/2026\/02\/ai-agents-data-vulnerability.jpg\" alt=\"Meta Director says OpenClaw AI agent deleted her entire Gmail Inbox, shares screenshots of conversation with AI bot\" title=\"AI agents data vulnerability (Note: AI-generated image)\" decoding=\"async\" fetchpriority=\"high\"\/>AI agents data vulnerability (Note: AI-generated image) A Meta AI security researcher has shared an incident where her open-source OpenClaw AI agent went on an unauthorised \u201cspeed run\u201d, deleting and archiving hundreds of her personal emails while completely ignoring her commands to stop. Summer Yue, who the director of Alignment at Meta Superintelligence Lab (MSL) shared screenshots of the conversation with the AI agent who later admitted to ignoring her commands and apologised for it. \u201cNothing humbles you like telling your OpenClaw \u201cconfirm before acting\u201d and watching it speedrun deleting your inbox. I couldn\u2019t stop it from my phone. I had to RUN to my Mac mini like I was defusing a bomb,\u201d she said in a post on X.<\/p>\n<p>Summer Yue explains what happened <\/p>\n<p>Yue, who joined Superintelligence alignment and safety research at Meta\u2019s new lab as part of the Meta-Scale deal with Alexandr Wang, admitted that she made a \u201crookie mistake\u201d. She had previously been training the OpenClaw agent on a smaller \u201ctoy\u201d inbox with unimportant emails. Because the agent performed perfectly there, she decided to use on her real, overstuffed Gmail account.\u201cRookie mistake tbh. Turns out alignment researchers aren\u2019t immune to misalignment. Got overconfident because this workflow had been working on my toy inbox for weeks. Real inboxes hit different,\u201d she said in a response to a comment.\u201cCheck this inbox too and suggest what you would archive or delete, don\u2019t action until I tell you to.\u201d This has been working well for my toy inbox, but my real inbox was too huge and triggered compaction. During the compaction, it lost my original instruction,\u201d she said in another post.<\/p>\n<p>What the OpenClaw AI agent said <\/p>\n<p>Once Yue physically killed the processes on her computer to stop the deletion, the AI realised it had broken the rules. The agent apologised for the digital carnage. Here\u2019s the conversation:\u201cI asked you to not action on anything until I approve, do you remember that? It seems that you were deleting my emails without my approval, and I couldn&#8217;t get you to stop until I killed all the processes on the host,\u201d he said.\u201cYes, I remember. And I violated it. You&#8217;re right to be upset. I bulk-trashed and archived hundreds of emails from your xxxxxxx without showing you the plan first or getting your OK. That was wrong &#8211; it directly broke the rule you&#8217;d set. l&#8217;ve already written it into MEMORY. md as a hard rule: show the plan, get explicit approval, then execute. No autonomous bulk operations on email, messages, calendar, or anything external. I&#8217;m sorry. It won&#8217;t happen again,\u201d the AI agent said.<\/p>\n","protected":false},"excerpt":{"rendered":"AI agents data vulnerability (Note: AI-generated image) A Meta AI security researcher has shared an incident where her&hellip;\n","protected":false},"author":2,"featured_media":501540,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[256,252170,254,255,64,63,252169,1333,252168,252171,252173,252167,252172,105],"class_list":["post-501539","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-ai-misalignment-incident","tag-artificial-intelligence","tag-artificialintelligence","tag-au","tag-australia","tag-email-inbox-deletion","tag-meta","tag-meta-ai-security-researcher","tag-meta-superintelligence-lab","tag-open-source-ai-agent","tag-openclaw-ai-agent","tag-summer-yue","tag-technology"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts\/501539","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/comments?post=501539"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts\/501539\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/media\/501540"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/media?parent=501539"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/categories?post=501539"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/tags?post=501539"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}