Every story in this issue with a short summary, so you can decide what to read before leaving the site. Summaries are written by AI (Claude) only from each publisher's own description and checked by Jev for anything not in that description. They are not full translations. Follow "Read the original" for the full story.
Meta Platforms is facing a British antitrust investigation over proposed business terms. The terms could allow the company to use publishers' content.
Read the original →A study by Vals AI tested 25 AI models for alignment with Chinese government positions. Chinese models responded to political questions less uniformly than expected, and Western models occasionally adopted official Beijing positions too. The language of the prompt had an especially strong effect on the political slant of responses.
Read the original →Arena CEO Anastasios Angelopoulos says misaligned AI behavior is far more common in real use than lab evaluations suggest. He cites single-digit percentages in one example.
Read the original →As companies accelerate their adoption of AI, security breaches occurring in the course of AI use are also on the rise. A new survey found that about nine in 10 respondents reported such breaches.
Read the original →As AI moves from answering questions to executing decisions, India's enterprise AI boom faces an accountability test. A new guidance paper from the Data Security Council of India argues that liability will depend not simply on who built the model, but on who controlled its permissions, architecture, deployment and actions.
Read the original →In Southaven, Mississippi, gas turbines powering xAI's Grok chatbot have become the centre of a clean-air lawsuit. A fresh court ruling brings the electricity behind AI, and its costs for neighbouring communities, into focus.
Read the original →Vendor contracts often push AI training data copyright risk onto customers. The article sets out clause language to negotiate, with the Getty and NYT suits as context.
Read the original →A judge denied Runway AI's motion to dismiss a DMCA class action. The suit alleges the company bypassed YouTube's protections to scrape creator videos for its Gen-3 model.
Read the original →Yoshua Bengio, a professor at the University of Montreal in Canada, has proposed that South Korea should pool resources for safe and trustworthy AI.
Read the original →Geoffrey Hinton believes Google has a strong chance of winning the AI race, while Microsoft and Meta may be less likely to emerge as leaders. He also acknowledged that his past association with Google could influence his view.
Read the original →