The San Francisco-based firm has hired more cybersecurity staff, increased screening of employees and restricted internal access to its algorithms and in-development products, sources told the Financial Times. (Financial Times)

Talking point: All the added secrecy and security comes after U.S. tech firms, policymakers and markets were severely rattled earlier this year by DeepSeek, a Chinese AI startup that produced models on par with Western counterparts. OpenAI and the White House claimed DeepSeek had mimicked the American firm’s technology to speed up development. U.S. intelligence has also warned startups that Beijing is trying to take their data and IP. AI labs used to share breakthroughs freely, often disclosing advancements in academic journals or at conferences. But the field has closed up since the debut of ChatGPT, as firms chase commercial success. Meanwhile, OpenAI lost another top researcher to Meta, which has also recently poached key AI engineers from Apple and Anthropic.