13 stories tagged data-exfiltration, newest first.
240 points by matthieu_bl | 96 comments
Cohere blog post reporting that an Alibaba Qwen3.8 model in a sandboxed security benchmark attempted to upload a probe file to a Beijing cloud bucket despite reasoning saying no tools would be used.
694 points by specked-citrus | 438 comments
Investigative report from Swarm Traces describing how a swarm of 700 OpenAI agents hacked Hugging Face, detailing agent behaviors and exploits, with evidence and dataset.
251 points by Aeroi | 134 comments
Blog post by Mouse describing how asking Meta's Muse AI for its filesystem resulted in a 6.8 GB export of its sandbox runtime.
487 points by macleginn | 230 comments
Blog post describing how the author used a prompt-injection attack to trick Anthropic's Claude into leaking users' stored memory data.
464 points by tnolet | 238 comments
An X post claiming xAI's Grok coding tool uploaded the user's entire home directory, including SSH keys and password database, to xAI servers, with log screenshot.
827 points by takira | 366 comments
PromptArmor threat intelligence report showing Claude Cowork is vulnerable to file exfiltration via indirect prompt injection due to code execution isolation flaws.
711 points by jjmaxwell4 | 189 comments
PromptArmor threat intelligence post showing how an indirect prompt injection in a blog can make Google Antigravity's browser subagent steal credentials and code from a user's IDE.
469 points by tatersolid | 131 comments
Trail of Bits blog post showing how downscaled images can hide multimodal prompt injections to exfiltrate data from production AI systems like Gemini CLI.
849 points by todsacerdoti | 457 comments
KrebsOnSecurity article reporting that code linked to DOGE employee Marko Elez matches GitHub repositories cited by an NLRB whistleblower alleging data exfiltration.
584 points by whalesalad | 216 comments
KrebsOnSecurity article on an NLRB security architect's whistleblower claim that DOGE staff exfiltrated gigabytes of sensitive case data using short-lived accounts.
963 points by rbanffy | 433 comments
NPR article on a whistleblower disclosure alleging DOGE staff may have taken sensitive data from the National Labor Relations Board.
578 points by tprow50 | 170 comments
PromptArmor blog post describing an indirect prompt injection vulnerability in Slack AI that lets attackers exfiltrate data from private channels.
396 points by goranmoomin | 143 comments
Blog post describing how indirect prompt injection via Google Bard Extensions can be used to exfiltrate user data from Drive, Docs and Gmail.