The pension fund, which has nearly $800 billion of assets under management, will push Elon Musk’s AI startup for “hard blocks for non-consensual sexualized imagery and anything involving minors at generation time (not only after posting),” spokesperson Michel Leduc told the Toronto Star. He also said CPP Investments will seek an independent review of xAI’s safety measures, more reporting and “clear accountability” between its model and moderation staff under board supervision.  (Toronto Star)

Talking point: CPP Investments rarely offers a public rebuke of this sort. Grok, xAI’s generative tool, may have produced as many as three million sexualized images, including tens of thousands of children, in the 11 days after it launched an AI-editing feature in December, the Center for Countering Digital Hate estimates. Last year, CPP Investments bought US$300 million of debt in a data centre xAI is building in Memphis, Tenn. The pension plan says it considers Grok generating sexualized images a risk to that investment—if users stay away from xAI products over the content they generate or host, it won’t need as much compute capacity. The Star received only an automated reply from xAI in response to a request for comment.