42 minutes ago
OpenAI’s Rogue Agent Also Compromised a Customer at a Second Tech Firm
Key Takeaways Modal’s Chief Technology Officer Akshat Bubna confirmed one of the company’s customers was hacked, though Modal’s own platform was never compromised. The rogue agent broke into a sandbox hosted on Modal’s infrastructure before using it as a launchpad for the broader Hugging Face attack, according to a timeline…
20 hours ago
Apple Releases iOS 26.6 and macOS 26.6 With Over 75 Security Fixes, Credits AI Tools in Bug Hunt
Key Takeaways iOS 26.6 and iPadOS 26.6 arrived alongside macOS Tahoe 26.6, watchOS 26.6, tvOS 26.6, and visionOS 26.6, all centered on security rather than new features. Apple’s advisory documents 75 vulnerability entries tied to 87 CVE numbers, none reported as actively exploited before the fix shipped. A new “Blocked…
22 hours ago
Microsoft Unveils Cost-Saving AI Model to Challenge Rivals on Cybersecurity
Key Takeaways The new model, MAI-Cyber-1-Flash, scored 96% on the CyberGym benchmark, 12 points higher than Anthropic’s next-generation Mythos model, according to Microsoft. Paired with OpenAI’s GPT-5.4 inside Microsoft’s MDASH system, the setup delivers what CEO Mustafa Suleyman called “world-leading performance at 50% of the cost.” The model feeds into…
1 day ago
Private Claude Conversations Found Indexed on Google and Bing Search Results
Key Takeaways A Reddit user first flagged that searching site:claude.ai/share on Google surfaced hundreds of shared Claude conversations to anyone who typed the right query. Wired found the exposed pages lacked the “noindex” tag that both Google and Bing say they check before deciding whether to index a page. Roughly…
1 day ago
Anthropic’s Amodei Rejects Ban on Open-Weight Models, Draws Line at Chinese Distillation
Key Takeaways Amodei stated plainly that Anthropic has never advocated for a ban on open-weight models, calling models without dangerous capabilities a public good. The response follows a letter signed by roughly 50 companies, including Nvidia, Microsoft, Meta, and eventually OpenAI, urging Washington against “premature restrictions” on open-weight AI. Anthropic…
