[MODEL] Why am I paying extra tokens just to correct outdated answers from Claude?

### Preflight Checklist

- [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Amodel) for similar behavior reports
- [x] This report does NOT contain sensitive information (API keys, passwords, etc.)

### Type of Behavior Issue

Claude modified files I didn't ask it to modify

### What You Asked Claude to Do

Asked a normal current-information question. Not breaking news.

### What Claude Actually Did

Gave an outdated answer confidently, then I had to use web/research and spend more usage just to get the correct answer.

### Expected Behavior

Claude should either verify potentially outdated information automatically or not make users spend extra usage to correct stale built-in knowledge.

### Files Affected

```shell

```

### Permission Mode

Accept Edits was ON (auto-accepting changes)

### Can You Reproduce This?

Yes, every time with the same prompt

### Steps to Reproduce

_No response_

### Claude Model

Opus

### Relevant Conversation

```markdown

```

### Impact

Critical - Data loss or corrupted project

### Claude Code Version

2.1.268

### Platform

Anthropic API

### Additional Context

Are you planning to do anything about this? Usage/token limits already feel tighter, so paying extra just to correct outdated answers is getting pretty frustrating.

No files, tests, or conversation transcript are provided. Start by obtaining the exact prompt, outdated answer, and research context, then reproduce it with Claude Opus 2.1.268 through the Anthropic API; done would require an agreed behavior for handling potentially stale answers and the associated usage impact.

Written by the indexing model from the issue text.