A collaboration between IBM Research and UC Berkeley has led to significant insights into the failures of agentic systems in IT automation, utilizing the ITBench benchmark and the MAST taxonomy.
Research reveals that AI chatbots often provide verbose and inaccurate responses to government service inquiries, raising concerns about their reliability.
Recent findings reveal that passwords generated by AI tools like Claude, ChatGPT, and Gemini may appear complex but are fundamentally weak and predictable.
Anthropic has launched Sonnet 4.6, an upgraded model that excels in coding and computer usage while maintaining a focus on safety.
A recent webinar highlighted the challenges of cloud forensics and the necessity of modern techniques to investigate breaches effectively.
A recent SmartLoader attack has been identified, leveraging a trojanized version of the Oura MCP server to deploy the StealC infostealer, targeting sensitive user data.
Following significant backlash from Hollywood, ByteDance is scrambling to implement safeguards for its AI tool Seedance 2.0, which has been accused of infringing on iconic characters.
A KPMG partner in Australia faced a AU$10,000 fine after using AI to complete an internal training exam on artificial intelligence, highlighting the challenges firms face in regulating AI usage.
Keir Starmer announces a swift approach to implement age limits on social media, targeting harmful content and addictive features.
A new agent skill enables coding agents to create production-ready CUDA kernels, enhancing performance for specialized tasks in AI models.
