AI

Searchable's analysis of 11,000 ChatGPT ads found 33% unrelated to the conversation, with 68% in chats without purchase intent.

A tutorial on auditing preference biases and fine-tuning language models with DPO on Anthropic HH-RLHF using TRL and LoRA.

A bookseller's hidden Airtag tracked bulk rare book orders to an Amazon AI training facility, confirming destructive scanning.

Zhipu's GLM-5.3 leads one cyber benchmark but trails on exploit tests. A deep dive into the numbers.

DeepSeek releases DeepSeek Harness v0.1 in developer preview, an MIT-licensed agent harness where everything is a plugin, built on the Cordis kernel.

OpenAI and Anthropic cut prices as Chinese rivals like DeepSeek and Moonshot gain ground, with token prices down nearly 25% since mid-July.

Google DeepMind's WeatherNext model improves hurricane forecasting by a day, marking a major shift in climate predictive tech.

An MIT-led study shows why health AI interfaces must adapt to user expertise to prevent costly automation bias.

Microsoft released code-testing-generator, an MIT-licensed agent that outperforms stock Copilot by planning and verifying unit tests.

Research presented at Black Hat reveals thousands of servers remain vulnerable to critical exploits in baseboard management controllers.

Auterion is supplying AI-powered strike kits to equip 50,000 Ukrainian Shrike drones with autonomous terminal guidance.

Article 50 of the EU AI Act has entered into force, mandating transparency for AI providers and deployers across the bloc.