AI News, 13 November 2026

Today · Fri 13 Nov

186 Fake AI Domains Redirect Users to ClickFix Lures, Browser Malware, and Search Hijackers

AI-generated summary Cybersecurity researchers discovered 186 lookalike domains impersonating artificial intelligence brands to distribute ClickFix lures, browser-hijacking malware, and search-hijacking extensions. Island reported that the campaign includes pathways delivering PowerShell commands, the Net Guard app, and extensions such as Quick Search and GPT for Chrome.

Apple Derives Theoretical Limits of KL-Regularized Language Model Alignment

AI-generated summary Apple research has characterized the information-theoretic limits of KL-regularized language model alignment by deriving the maximum reward improvement achievable under a given KL-divergence budget. The study examines reinforcement learning alignment and best-of-N alignment to establish bounds on how much a model's outputs can be steered toward human preferences before capability degradation occurs.

AI chatbots are recommending fake retail websites, study finds

Shopping basket with an 8-bit effect.

AI-generated summary A study by Podqi testing ChatGPT, Gemini, and Perplexity found that chatbots displayed risky URLs in up to 15 percent of conversations. The research revealed that Perplexity showed risky sites in 15 percent of answers, while ChatGPT reached 8 percent and Google's Gemini showed them in 4 percent of answers.

Also covered by Artificially Nerdy