AI models are rapidly improving at exploiting security vulnerabilities. According to a new study, their offensive cyber capability has been doubling every 5.7 months since 2024, with Opus 4.6 and GPT-5.3 Codex now solving tasks that take human experts about three hours.

Reinforcement learning hits a wall with reasoning models because every token gets the same reward. A new algorithm from Alibaba's Qwen team fixes this by weighting each step based on how much it shapes what comes next, doubling the length of thought processes in the process.

AI tools can speed up journalism until they backfire. Two recent cases show what happens when writers don't understand how their AI tools work: copied passages and made-up quotes.

Can’t stop the slop.