What happened in this Tech story?

Anthropic says an unreleased AI model turned security flaws into working proof-of-concept exploit chains across five software platforms, demonstrating a lab-built technique rather than evidence of an attack in the wild.

What is a key fact in this story? (1)

**Anthropic** said in its latest report that an unreleased AI model found security flaws in Chrome, Safari, Android, Windows and Linux.

What is a key fact in this story? (2)

**Anthropic's unreleased model** chained those flaws into working proof-of-concept exploits that were built in the lab after the model was prompted and had not been seen in any real attack.

📎 SkimNews has covered Anthropic 565+ times · see the file →

Anthropic Model Chained Flaws Into Working Exploits — SkimNews

By The Hacker News · Summarized & edited by Kapil Samant · 2026-10-28

Get the Tech newsletter

Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.

Anthropic said in its latest report that an unreleased AI model found security flaws in Chrome, Safari, Android, Windows and Linux.
Anthropic's unreleased model chained those flaws into working proof-of-concept exploits that were built in the lab after the model was prompted and had not been seen in any real attack.

Why it matters: For Anthropic, the claim is stronger than a bare vulnerability list: the model produced working exploit chains across five named platforms—Chrome, Safari, Android, Windows and Linux. The material boundary is equally clear: the exploits were prompt-built in a lab, and no real-world attack was observed.