MIT Technology Review · AIMay 21, 2026, 22:30 UTC+8

Anthropic’s Code with Claude Showed Off Coding’s Future—Whether You Like It or Not

Anthropic’s Code with Claude showed off coding’s future—whether you like it or not

Anthropic used its two-day Code with Claude event in London to show Claude Code automation, with nearly half the room saying they shipped a pull request fully written by Claude in the past week, and many keeping their hands raised when asked whether they had shipped it without reading the code.

Why it matters: HKR-H/K/R all pass: the MIT Tech Review piece has a strong Claude Code hook, a concrete developer-behavior number, and clear resonance for programmers. It is not a model release or major product launch, so it stays in the 78–84 band.

Read the original ↗Export Markdown

Take

Anthropic’s sharpest signal is not Claude Code writing PRs; it is developers admitting they shipped Claude-written PRs unread. Nearly half the room said Claude wrote a full PR in the prior week. Most hands stayed up when asked if they skipped reading the code. That crosses from Copilot-style acceleration into an org process where tests, retries, and green checks replace human inspection.

The concrete hook is strong: a two-day London Code with Claude event, with Anthropic pointing to Claude 4.6 and 4.7 and saying the default becomes “Claude prompts itself.” I don’t buy the victory-lap framing. The article gives no risk class, test coverage, rollback rate, or production severity for those unread PRs. The sample is also a Claude fan room. The scary part is not faster code; it is review becoming theater.

Source: MIT Technology Review · AI · technologyreview.com
Published: May 21, 2026, 22:30 UTC+8 · May 21, 2026