AI Cheating happening more and is hard to verify
10 hours ago
- OpenAI launches 'Presence' consulting service to deploy AI agents, suggesting margins lie in implementation as models commoditize.
- A Windows stealer malware targets over 300 apps and uses an AI profiler to maximize profits for criminals.
- Chinese open-source AI models are now competitive globally, challenging assumptions about superiority.
- AI trust issues persist as verification becomes difficult; even with 'trust but verify' approaches.
- Nvidia introduces the Vera Rubin platform to optimize token emission for AI factories.
- Russian actors pose as Signal support to launch phishing attacks, alongside other security incidents like Iranian propaganda takedowns.
- Microsoft fails to patch on-prem SharePoint, which is now under zero-day attack; China upgrades surveillance tools.
- DEF CON's Franklin project enlists hackers to harden critical infrastructure, with voting village success expanding to entire conference.
- EQT acquires majority stake in Swiss cybersecurity firm Acronis at a $3.5B+ valuation.
- Ten years after the first corporate ransomware, infosec remains a stable career as attacks continue.
- GNOME offers a Windows-like taskbar option without extensions via the new 'Simple-taskbar'.
- Debian releases final version for x86-32 architecture, while new versions 13.6 and 12.15 are out.
- Joomla website extensions iCagenda and Balbooa Forms have critical bugs exploited with perfect 10 CVSS scores.
- A new X11 server called Frame is implemented directly in assembly, joining other alternatives.
- Cinnamon 6.8 adds optional Wayland support, while KDE Plasma 6.6.6 arrives, and Wayland becomes mandatory in 6.8.