# Amazon Quick adds Smart Approvals

> AWS describes review for eligible agent actions.

By BIGCHANGE.AI Editorial

Published: 2026-11-05T13:39:37.968Z
Updated: 2026-11-05T13:39:37.968Z
Canonical: https://bigchange.ai/blog/amazon-quick-smart-approvals

![Conceptual illustration of a person reviewing blank papers beside a key.](https://bigchange.ai/api/media/file/action-review-concept.png)
AI-generated conceptual illustration by BIG CHANGE.

AWS [launched Smart Approvals](https://aws.amazon.com/about-aws/whats-new/2026/10/quick-smart-approvals-agentic-actions/) in Amazon Quick on October 28; a [November 4 post](https://aws.amazon.com/blogs/machine-learning/smart-approvals-in-amazon-quick-human-oversight-where-it-matters/) explains its operation.

## The big change

- Existing denials and mandatory human approvals still apply.
- Single-use approval covers one action and its exact parameters; failures or timeouts return to the user.
- A classifier can make mistakes; mandatory review should be configured in policy.

## How the decision works

According to the [Quick guide](https://docs.aws.amazon.com/quick/latest/userguide/smart-approvals.html), automated review applies when the connector lets the user choose and the user selects “Ask each time.” An action set to always ask still requires a person; an always-allow action follows that setting directly. Smart Approvals is on by default, but deny-by-default profiles require explicit permission. Users can turn it off across conversations and scheduled runs. It is available in desktop and mobile apps, not the web version.

A separate classifier considers the request and conversation, potentially approving low-risk, request-consistent actions to reduce routine prompts; consequential or uncertain actions return to the user. In AWS’s example, a retrieved note adds a large mailing list to two requested recipients. The changed send is designed to return for review, not guaranteed to do so.

The broader issue is indirect prompt injection: [NIST](https://csrc.nist.gov/glossary/term/prompt_injection) defines prompt injection as an attack involving untrusted input and a higher-trust prompt. A document can be useful task material while also containing instructions that should not become the user’s authorization. The example illustrates that risk, not measured detection.

## What the logs establish

AWS says an approval-related CloudTrail event does not prove execution. Investigators can compare it with conversation logs and destination-system records, where available, to reconstruct what happened.

## Sources

- [AWS launch](https://aws.amazon.com/about-aws/whats-new/2026/10/quick-smart-approvals-agentic-actions/) — October 28.
- [AWS operational post](https://aws.amazon.com/blogs/machine-learning/smart-approvals-in-amazon-quick-human-oversight-where-it-matters/) — November 4.
- [AWS Quick guide](https://docs.aws.amazon.com/quick/latest/userguide/smart-approvals.html) — Eligibility and settings.
- [NIST: prompt injection](https://csrc.nist.gov/glossary/term/prompt_injection) — General definition.