The European Parliament has a problem with AI, and its answer is more AI. Lawmakers keep using public chatbots to help write legislation. So the institution is building a sanctioned platform to bring that habit under control. The internal tool, called EPGenAI Hub, will give MEPs and staff access to models from Meta, OpenAI, Anthropic […]<br /> This story continues at The Next Web [...]
Researchers at the UK AI Security Institute used psychometric methods to show that popular safety benchmarks for language models don't measure one consistent trait. Blanket blocking of requests can artificially inflate a safety score even as the model gets less useful day to day. The study also offers a method for catching models that act more cautious during tests than they do in normal use. [...]