Skip to content
IAM Insights

Explainers · Glossary

Preventative controls

preventative control · preventive controls

Controls that block a disallowed action before it happens (for example cloud guardrail policies), as opposed to detecting it afterwards.

In our briefings

The claim each of the newest briefings on Preventative controls would have you leave with.

  1. Ramani says Block's working setup rests on a third-party attack-surface tool that product security already ran. Another application pulls DNS records from Block's registrars into it, so the tool sees both the records and the AWS, GCP and Oracle accounts and acts as the single source of truth. When a record becomes open to takeover, the tool sends a webhook and Block's routing opens a ticket by type: hosted-zone cases go to platform security, which has registrar access, and other cases, such as CNAME, MX or Elastic IP records, go to the team that owns the account in Block's application catalog. A ticket closed while the record is still exposed reopens. The system runs on Lambda, Step Functions and API Gateway and keeps no state of its own, since the tool holds it. He puts DNS takeovers in Block's bug bounty program at "roughly around 200" in 2022 and "almost zero" by 2024, without saying whether he means reports or findings.

    A Block engineer says routing dangling DNS records to their owners cut takeover bug bounties to almost zeroBI-FIND-2026-2231

  2. ENISA's March 2024 report says the main attack on remote identity proofing has moved behind the camera. It calls digital injection attacks using deepfakes "the predominant type of attack", with presentation attacks such as masks and replayed screens held up to the lens used "to a far lesser extent", and warns that a similar surge in deepfake presentation attacks "can occur anytime soon". The number it prints for the shift is iProov's: the vendor's operations centre saw injection incidents in 2022 at about five times the rate of presentation attacks. Its other attack figure concerns documents: in some Member States, it says, document attacks rose significantly within the last year, with about 9 in 10 identity spoofing attempts relating to identity documents, and it names no source. Presentation attack detection has a revised test standard, ISO/IEC 30107-3:2023. For injection detection ENISA found "no globally accepted standards", with CEN TC 224/WG 18 drafting one, and it expects more than five years of research before the candidate countermeasures are validated. That CEN document came out after the report, as technical specification CEN/TS 18099 in December 2024.

    In March 2024 ENISA found deepfake injection leading attacks on ID proofing, with no agreed standard to detect itBI-FIND-2024-0092

Cited in 2 briefings

Every briefing that discusses this term, newest first.

  1. A Block engineer says routing dangling DNS records to their owners cut takeover bug bounties to almost zero

    Podcast · · Cloud Security Podcast · How Block Solved DNS Vulnerabilities · Oct 13, 2026

  2. In March 2024 ENISA found deepfake injection leading attacks on ID proofing, with no agreed standard to detect it

    Article · · ENISA · Evgenia Nikolouzou et al. · Mar 12, 2024