TA0009Tactic 12 of 15

Collectionin cloud and AI environments

How do attackers collect data from cloud storage?

ATT&CK v20.1Updated Official tactic page
Collection technique map17 techniques in Collection, 5 of them apply to cloud platforms. Highlighted: T1530, T1213, T1213.003, T1119.T1119T1213T1530TA000917 TECHNIQUES
Key for cloud (this guide)Applies to cloud platformsOther platforms
17
Techniques
24
Sub-techniques
5
Apply to cloud platforms
4
Defensive controls
2
Detection signals

What the tactic means

Collection covers techniques for locating and gathering the data an adversary is after — documents, databases, code, messages — usually as preparation for exfiltration or extortion. In cloud environments it frequently looks like ordinary authorized reads by an identity that should not be making them.

The adversary is trying to gather data of interest to their goal.

Collection consists of techniques adversaries may use to gather information and the sources information is collected from that are relevant to following through on the adversary's objectives. Frequently, the next goal after collecting data is to either steal (exfiltrate) the data or to use the data to gain more information about the target environment. Common target sources include various drive types, browsers, audio, video, and email. Common collection methods include capturing screenshots and keyboard input.

MITRE ATT&CK TA0009 description

Why it matters: cloud, AWS and AI

Cloud environments

Cloud data is reachable through APIs; collection is often just authorized reads by the wrong party. Without object-level logging it may be impossible to determine what was read.

AWS specifics

S3 GetObject, DynamoDB Scan, RDS snapshot creation, Athena queries over data lakes and reading code repositories are typical collection activities.

AI & workloads

AI assistants connected to broad document sets can collect on an adversary’s behalf; ATLAS tracks data collection from AI services.

AWS API calls to know
  • s3:GetObjectRead objects (data event).
  • dynamodb:ScanRead whole tables.

What the attacker is trying to achieve

  1. 01Locate and gather valuable data
  2. 02Collect at scale without triggering alerts

Techniques that matter most in cloud

What it needs, and what breaks it

Attacker needs

  • Read permissions on data stores

Defenders control

  • Least-privilege data access
  • KMS key policies separating decrypt
  • Data events on sensitive stores
  • Authorization-aware retrieval for AI assistants

Detection signals and telemetry

Signal
  • Bulk object reads by one identity

    S3 data eventsGuardDuty S3 Protection
  • Table scans by unusual principals

    CloudTrail data events

For leadership

What this means for the business

Whether an incident is "potential exposure" or "confirmed exposure of N records" depends on logging configured before the incident.

Common misconceptions

Myth

“CloudTrail shows every S3 read.”

Reality

Object-level reads are data events and are not logged unless enabled.

Practical security checklist

0 of 3 done

Saved only in this browser. Nothing is sent to Securitain.

Related attack scenarios

Where Securitain fits

  • Securitain can assess todayDetection of identities with sensitive permissions such as broad S3 access, KMS decrypt or Secrets Manager reads. Learn more
  • Securitain can assess todayPublic and cross-account exposure checks on S3, KMS, SQS, SNS and Secrets Manager resource policies. Learn more

Official sources and references

MITRE ATT&CK® and ATT&CK® are registered trademarks of The MITRE Corporation. ATT&CK data © 2026 The MITRE Corporation. This work is reproduced and distributed with the permission of The MITRE Corporation (ATT&CK v20.1). Securitain is not affiliated with or endorsed by MITRE. Explanations, scenarios and control mappings on this page are Securitain’s own analysis. Sources and licensing

Stay in the loop

Cloud security insights, early access & compliance updates

Weekly InsightsEarly AccessExpert Content
No spam, everUnsubscribe anytimeWeekly updates
Chat with Securitain