Protein excess tips the balance of cellular health
By Robert Booth
Artificial intelligence is becoming a common writing assistant, but is it also changing what we mean to say? A new study from researchers at Oxford and Potsdam universities found that several popular AI models subtly rewrote users’ drafts on controversial topics such as abortion, climate change, religion, and politics. In some cases, the changes went far beyond improving grammar, completely reversing the original message or adding political framing the writer never intended.
The researchers warn that these small edits, repeated across millions of posts, could gradually influence public opinion without users even noticing. Different AI systems showed different political leanings, but the broader concern is the same: when AI rewrites our words, whose voice are we really hearing? If AI is becoming a gatekeeper of online communication, this research raises important questions about trust, bias, and the future of human expression.