AI security tools help Firefox uncover and fix hundreds of vulnerabilities, shifting the balance against zero-day exploits.
Anthropic is investigating unauthorized access to its Mythos cybersecurity model by a Discord group, raising fresh questions about control of powerful AI hacking tools.
Claude Code 2.1.88 leak exposed 512,000 lines via npm error, fueling supply chain risks and typosquatting attacks.
LLM Poisoning [1/3] - Lire les pensées d'un Transformer
A zero-day vulnerability in the Linux kernel was discovered, utilizing OpenAI's o3 model. This finding, designated as CVE-2025-37899, marks a significant advancement in AI-assisted vulnerability research.
Microsoft has denied claims that it uses Microsoft 365 apps (including Word, Excel, and PowerPoint) to collect data to train the company's artificial intelligence (AI) models.
A week after falsely claiming Kamala Harris’ campaign used AI to inflate crowd sizes, Trump is now posting AI-generated photos of Taylor Swift fans endorsing him.
Microsoft Copilot for Security is generally available April 1, 2024, with new capabilities to help protect and govern AI use.
Les Etats-membres de l'UE et le Parlement européen se sont accordés sur un texte pour encadrer l'utilisation et le développement de l'intelligence artificielle. Une législation inédite à l'échelle mondiale.
