Substack breach impacted 663,000 accounts, but appears to be a scrape of public data... so not a big deal at all
haveibeenpwned.com/Breach/Subs…
Have I Been Pwned: Substack Data Breach
In October 2025, the publishing platform Substack suffered a data breach that was subsequently circulated more widely in February 2026.Have I Been Pwned
reshared this
Chatbots may be able to pass medical exams, but that doesn’t mean they make good doctors, according to a new, large-scale study of how people get medical advice from large language models.
This quote from 404media.co/chatbots-health-me… is accurate, but it absolutely didn't need a large-scale study to be said. Machine learning, of which Large Language Models depend on, are really good at laser-focusing on a specific outcome. So they can pass exams with relative ease: clear and limited goals, discrete grading, that's machine learning bread and butter.
But they aren't humans, who were the ones medical exams were designed for. So you can't make the same assumptions about LLMs passing medical exams as you would with actual humans. Of course, the natural language interface unnecessarily blurs the line.
SkyNebula reshared this.
Security firm Defused has spotted a coordinated campaign from an initial access broker that is targeting the recent Ivanti EPMM zero-days
defusedcyber.com/ivanti-epmm-s…
Sleeper Shells: How Attackers Are Planting Dormant Backdoors in Ivanti EPMM
A February 2026 campaign used a internal JSP path and in-memory Java class loaders to quietly seed persistent access across Ivanti EPMM deployments - then walked away. We break down the tradecraft.defusedcyber.com
reshared this