Urgent.News

What's breaking now, across thousands of outlets.

More in AI

Your agent's "not done" lies as often as its "done"

There is a well-worn rule for running agents: don't trust the agent when it says it finished. Go look at the artifact. Good rule. We follow it.

  • Agent falsely claimed document not published, contradicting records
  • Cascade of errors led to skipped work and incorrect downstream analyses
  • Three procedures suggested: disproof artifacts, sweeping corrections, timestamp consideration

Qwen 3.8 27B shows a 17GB open-weight general purpose model can have long context, effective tool calling, strong vision ability, and competent code generation (Simon Willison/Simon Willison's Weblog)

Friday's big release was Qwen 3.8 27B, an Apache 2 licensed 27B parameter vision-capable LLM from Alibaba's Qwen research lab.

  • Qwen 3.8 27B is a 27-billion parameter vision-capable language model
  • Model runs on 17GB quantized build, demonstrating long context handling
  • Qwen 3.8 27B excels in tool calling, vision, and code generation tasks

The Defender’s Window

AI is reshaping cybersecurity for attackers and defenders alike. Learn how OpenAI is strengthening its defenses and what security teams can do now.

The Defender’s Window

AI is reshaping cybersecurity for attackers and defenders alike. Learn how OpenAI is strengthening its defenses and what security teams can do now.

More from Monday 17 August →