Poisoning AI models might be way easier than previously thought if an Anthropic study is anything to go on. Researchers at the US AI firm, working with the UK AI Security Institute, Alan Turing ...