Urgent.News

What's breaking now, across thousands of outlets.

AI

When the model answers in the language of the pasted text

A short request in one language, pasted material in another: Claude sometimes answers in the wrong one. What we measured, and a check that catches it. Mostly because the pasted text outweighs the request. When someone writes one short line in Bulgarian and pastes a long English build log under it, the model reads far more English than Bulgarian, and when it sums up the paste, the summary often…

We haven't written up this one. Dev.to has the full story — the link below goes straight to it.

Read the original at dev.to →

More in AI

Your AI Agent Didn't Break the Rules. One of Your Rules Was Missing.

Anatomy of an autonomy bug: when two valid decision paths create one invalid outcome. Part 1 — For everyone The thing about autonomous agents nobody tells you Building an autonomous agent is a bit…

  • Sentinel AI agent spent tokens without trigger on Oct 7 and Oct 8
  • Two decision paths allowed token spending without reason
  • Architecture lacked communication between two functioning checks

What Happens When Users Don't Behave as Expected?

In the previous article, we looked at why functional testing alone is not enough for AI applications. Traditional testing usually starts with a simple assumption: developers know how users are…

  • Users often behave unexpectedly, rephrasing requests or asking unrelated questions.
  • AI systems interpret natural language, leading to responses different from developer intentions.
  • Testing must go beyond predefined scenarios to evaluate unexpected user behavior.

Gemini & Claude: Google's AI Agent Gamble

The AI Tango: When Rivals Become Partners (or, My Morning with Gemini) My morning began with a deceptively simple request for Gemini.

  • Google and Anthropic form alliance, surprising tech world
  • Google's Gemini agents will support Anthropic's Claude 3.5 Sonnet
  • Strategy positions Google as foundational layer for AI agents

Portable AI Agent Memory: What Should Move When Users Switch Agents?

Your AI Agent Knows You. What Happens When You Leave? Imagine using an AI assistant for two years. It knows how you prefer reports to be structured.

  • AI agents store active conversation, task state, and user preferences during transitions.
  • Useful context should be preserved without granting operational authority to new agents.

More from Friday 9 October →