Claude Model Builds Exploits Across Major Browsers & OSes
Source: Dark Reading (opens in a new tab) · Jai Vijayan
Intel Summary
Dark Reading reports that an unreleased version of Anthropic's Claude model has demonstrated the capability to discover and chain multiple vulnerabilities across major platforms, including Chrome, Safari, Android, Windows, and Linux, according to claims by Anthropic.
Why It Matters
Autonomous vulnerability discovery and multi-step exploit chaining across core operating systems and web browsers signal advancing offensive AI capabilities, raising critical release-gating and defensive security considerations for frontier model developers and platform maintainers.
Organizations & Entities
Topics
Related Intelligence
- Report
Google launches Gemini 4 Carbon with a ‘significant leap’ in cybersecurity
Google has launched Gemini 4 Carbon, its latest frontier model claiming substantial improvements in cybersecurity, coding, and agentic workflows. According to Google, the model ranked first on the Artificial Analysis Intelligence Index (v5.0) benchmark, outpacing frontier models including Claude Opus 5.5 and GPT-6 Astra.
The Verge - Report
Mistral releases open weights for Mistral Large 4, providing sovereign control for the company’s ‘Le Chonk’ AI model
SiliconANGLE reports that Mistral AI has released the open weights for Mistral Large 4, nicknamed 'Le Chonk'. The natively multimodal model, originally previewed three weeks prior, features 1 trillion total parameters with 52 billion active parameters.
SiliconANGLE - Report
Google's AI search problem finally shows up in the earnings report, some investors believe
Google's stock fell up to 10 percent after quarterly earnings showed search revenue growth slowed to 12 percent, missing expectations despite cloud revenue doubling to $30.7 billion. Google acknowledged that its expanding AI search mode yields less revenue than traditional search and announced the international expansion of its AI mode advertising program.
The Decoder - Report
‘Nobody’s Ready’: AI-Found Bugs Spark a Corporate Security Spending Spree
Enterprise software startup Atomicwork directed its engineering team to halt standard work and focus entirely on cybersecurity following multiple incidents over a 48-hour period. Reported compromises included a breach at root certificate authority eMudhra, a security compromise involving open-source AI request routing tool LiteLLM, and a newly discovered vulnerability in PHP.
The Information