Researchers found that "Fable 5 would quietly reroute requests to a lesser model when asked to perform certain actions. Moreover, that restriction wasn't disclosed in the model's documentation." The model "was either refusing or degrading responses for tasks like training competing LLMs, debugging AI code and optimizing neural architecture. Researchers were bothered not only by that degradation but by Anthropic's lack of transparency about it. They were also concerned, of course, that they had burned tokens and money for a model that didn't do what they expected."Engadget: Anthropic backtracks on policy that 'sabotaged' researchers' work https://www.engadget.com/2192004/anthropic-walks-back-policy-sabotaging-research/ @Engadget #Anthropic #LLM
Related
Corporate: we can't afford good enough CI runners, #AWS is expensive.Also Corporate: it's now mandatory to burn as many ...
Corporate: we can't afford good enough CI runners, #AWS is expensive.Also Corporate: it's now mandatory to burn as many #Bedrock tokens as possible.#GenAI #LLM
This smartwatch packs a battery that could shame every major smartwatchHONOR's new watch might actually have the biggest...
This smartwatch packs a battery that could shame every major smartwatchHONOR's new watch might actually have the biggest battery in any major smartwatch.https://www.androidauthorit...
⚠️ DeepSeek nel mirino di Washington: i presunti legami militari spingono l’AI cinese a un passo dalla blacklist USA. #A...
⚠️ DeepSeek nel mirino di Washington: i presunti legami militari spingono l’AI cinese a un passo dalla blacklist USA. #AI #Geopolitica🔗 https://www.tomshw.it/hardware/deepseek-sfio...