A study of 535 real scanned receipts from the ICDAR-2019 SROIE dataset found that metamorphic testing — checking whether an LLM returns consistent totals when the same receipt is presented in differen…
A new signed symmetric quantization method eliminates clipping errors in low-bit AI models without the runtime penalty of asymmetric quantization, achieving up to 2.45 times faster inference on AMD EP…
Zscaler found that several high-end AI agents, including Llama3 and Gemini models, fell victim to indirect prompt injection traps that would fool few humans, while some lower-level LLMs performed bett…
Zscaler found that several major AI models, including Llama3 and Gemini variants, are vulnerable to indirect prompt injection traps, while others like Llama4-maverick and Gemini-3.1-pro proved safe. E…
A developer built a daily AI paper scanner using Ollama and Telegram, automating the retrieval and translation of HuggingFace Papers into Portuguese. The Python script fetches the day's top papers via…