Topic: proprietary models
-
China's Free AI Model Outperforms GPT-5 and Sonnet 4.5
Moonshot's new open-source AI model, Kimi K2 Thinking, claims to outperform top proprietary models like GPT-5 and Claude Sonnet 4.5 on key benchmarks including reasoning and information retrieval. The model is freely available, trained for just $4.6 million, and uses a Mixture-of-Experts architec...
Read More » -
Microsoft's AI guardrails bypassed with a single prompt
Modern AI safety systems are surprisingly fragile, as a single, carefully crafted prompt can often bypass established guardrails, raising urgent questions about long-term reliability. Researchers used a technique called GRPO Obliteration to steer AI models away from safety constraints by rewardin...
Read More » -
Apple's New AI Chatbot Launches Early, But There's a Catch
Apple has launched an internal AI chatbot named Asa for retail staff to enhance sales readiness and support ahead of new product launches. The tool reflects Apple's strategy of prioritizing internal AI applications over consumer-facing products, focusing on operational efficiency and employee emp...
Read More » -
Switzerland Unveils Open-Weight AI Model for Developers
Switzerland has launched Apertus, an open-weight AI model that provides a transparent and legally compliant alternative to proprietary systems like ChatGPT, aligning with EU copyright standards and ethical data practices. Apertus offers full access to its source code, training data, and documenta...
Read More » -
3 Warning Signs Your AI Model Is Secretly Poisoned
Model poisoning is a deliberate security threat where attackers embed hidden backdoors during training, which remain dormant until a specific trigger activates them, making detection difficult. Key indicators of a poisoned model include a sudden, illogical shift in attention when triggered, the t...
Read More »