Auto-mode does not yet convincingly fix prompt-injection risk for coding agents.

Added to korrents 29 Aug 2026 · How quotes work · Something wrong? Tell us

Where this collides

• Cannot both be true

  For agents built on Claude Fable 5.1, it is in practice safe for now to assume they will not get prompt injected.

  Both men are looking at the same defence — auto mode — one month apart, and answering the same practical question opposite ways. Mowshowitz says it is in practice safe to assume your agents will not get prompt injected even if you are asking for it; Willison says he really wants to believe it fixes prompt injection for coding agents and is just not there yet. Someone deciding whether to point an agent at untrusted input cannot follow both.

Shared subject first, then similar wording. They may agree or disagree.

• Prompt injection can be caught by watching the neurons that fire when it happens, so a defence no longer depends on the model reporting the attack. BCBoris Cherny27 Jul 2026prompt injection
• Prompt injection has no equivalent of parameterized queries: there is no reliable way to tell a language model which text is data and which is instructions. SWSimon Willison19 Mar 2026prompt injection
• Prompt injection is an extremely hard attack to defend against, and hardly anyone integrating AI is talking about it. THThePrimeagen22 Mar 2025prompt injection
• Prompt injection is no longer a live attack on the frontier model: it simply does not follow instructions it reads on the internet any more. BCBoris Cherny27 Jul 2026prompt injection
• Prompt injection is unsolved, but the latest models are post-trained to resist it, so it takes far more than 'ignore all previous instructions' now. PSPeter Steinberger12 Feb 2026prompt injection

On the map

Open the map on its own page