Skip to main content

🎙️ Live Dec 10: Why controls behind CMMC matter

  • blog
  • The Proposed 2027 SOC 2 Trust Services Criteria: Changes, Timeline, and How to Prepare

The Proposed 2027 SOC 2 Trust Services Criteria: Changes, Timeline, and How to Prepare

  • November 11, 2026
Author

Anna Fitzgerald

Senior Content Marketing Manager

Reviewer

Rob Gutierrez

Senior Cybersecurity and Compliance Manager, CISA, CCSK, CMMC RP

Every SOC 2 report issued in the last several years has been measured against the same yardstick: the AICPA's 2017 Trust Services Criteria, with points of focus refreshed in 2022. In late October 2026, the AICPA proposed the first substantive rewrite of those criteria since they were introduced.

The exposure draft of the proposed 2027 SOC 2 Trust Services Criteria consolidates overlapping criteria, reorganizes the common criteria, adds new criteria for resilience testing and data disposal, and works AI risk into the points of focus across much of the framework. The AICPA is taking public comment through January 15, 2027. If the criteria are finalized, they would apply to SOC 2 engagements beginning on or after December 31, 2027.

Nothing changes for audits you have scheduled today. But the proposal shows where auditor expectations are heading, and organizations that use AI in their products or operations have good reason to start reviewing their controls now. This guide covers what the draft changes, how the timeline works, and what service organizations can do to prepare.

Comparison of SOC 2 Trust Services Criteria counts by category, showing the proposed 2027 criteria reducing the total from 61 to 50.

Current SOC 2 criteria: the 2017 TSC with 2022 points of focus

Today's SOC 2 examinations use the 2017 Trust Services Criteria with revised points of focus from 2022, published by the AICPA in TSP section 100. Until the proposal is finalized and takes effect, these remain the criteria your auditor will test against.

The framework is organized into five Trust Services Criteria categories: Security, Availability, Processing Integrity, Confidentiality, and Privacy. Security is required in every SOC 2 report, and the other four are added based on the commitments you make to customers. Across all five categories there are 61 criteria, and 33 of them sit in the Security category, also called the common criteria.

Each criterion comes with points of focus, which describe characteristics an auditor may consider when evaluating whether your controls meet it. Points of focus are guidance for auditors. You design your own controls, and your auditor judges whether those controls meet the criteria.

The 2017 criteria were aligned with the 17 principles of the COSO 2013 internal control framework, which is why the first five common criteria series (CC1 through CC5) map to COSO's components. That alignment is one of the things the proposal changes.

What's changing in the proposed 2027 SOC 2 Trust Services Criteria?

The proposal cuts the total number of criteria from 61 to 50 while keeping the same five categories and the same outcomes-based approach. The AICPA describes the update as a significant modernization: duplicative concepts are consolidated, some criteria are removed, new criteria are added where more clarity or emphasis is needed, and existing criteria are reworded so each has one concise meaning.

Most of the reduction happens in the common criteria. The optional categories stay the same size or grow slightly.

Category 2017 criteria Proposed 2027 criteria What changes
Security (common criteria) 33 20 Consolidated and renumbered CC1 through CC8
Availability 3 4 Adds A1.4, resilience and recovery testing
Processing Integrity 5 5 Count unchanged; AI-related points of focus added
Confidentiality 2 3 Adds C1.3, disposal of data
Privacy 18 18 Count unchanged; P3.2 moves to P2.2
Total 61 50

A few other changes apply across the whole framework:

  • COSO references are removed. The proposed criteria are designed specifically for SOC 2 and SOC 3 engagements and no longer reference COSO’s 17 principles.
  • Governance language is broadened. "Those charged with governance" replaces "board of directors," which fits startups and private companies that don't have a formal board.
  • Criteria get short titles. Almost every criterion gains a title, which should make reports and control mappings easier to read.

For most organizations, a smaller and clearer set of criteria should mean less ambiguity during readiness and fewer debates with auditors over which criterion a control maps to. It also means your existing control mappings will need to be redone against the new numbering once the criteria are final.

How the SOC 2 common criteria would be reorganized

The common criteria shrink from nine series to eight. The standalone control activities series (CC5) goes away, and every series after it moves up one number.

2017 series 2017 count Proposed 2027 series Proposed count
CC1 Control environment 5 CC1 Control environment 2
CC2 Communication and information 3 CC2 Information and communication 2
CC3 Risk assessment 4 CC3 Risk assessment 3
CC4 Monitoring activities 2 CC4 Monitoring 1
CC5 Control activities 3 Removed 0
CC6 Logical and physical access controls 8 CC5 Logical and physical access 6
CC7 System operations 5 CC6 System operations 2
CC8 Change management 1 CC7 Change management 3
CC9 Risk mitigation 2 CC8 Vendor management 1

Access controls get new numbers. Logical and physical access, the series most SOC 2 programs spend the most time on, moves from CC6 to CC5. If your policies, control descriptions, or evidence requests reference CC6.1 or CC6.6, those references will need updating.

Change management grows. It goes from a single criterion to three, which signals more specific expectations for how changes to systems, infrastructure, and software are authorized, tested, and approved.

Risk mitigation becomes vendor management. The old CC9 combined business disruption and vendor risk. In the proposal, business disruption planning moves to the Availability category (A1.3), and CC8 focuses on vendors and other third parties.

Some 2017 criteria, including CC1.3, CC1.5, CC3.1, and CC5.2, have no direct successor in the draft. The AICPA has not published an official crosswalk from the 2017 criteria to the proposed ones, so treat any mapping you see, including this one, as an interpretation of the draft until the AICPA publishes official guidance.

How AI fits into the proposed SOC 2 criteria

The proposal addresses AI through new points of focus on existing criteria, so AI risk becomes part of how auditors evaluate controls you already have.

According to the AICPA, the new points of focus cover:

  • AI governance and oversight
  • Monitoring of AI models
  • Prompt manipulation
  • Data poisoning
  • Transparency of AI-generated outputs
  • Management of training data
  • Risks from automated decision-making

These points of focus show up broadly. One analysis of the draft counts AI language in 17 of the 50 proposed criteria, including 12 of the 20 common criteria. Examples include AI risk considerations in risk assessment (CC3.2), unsanctioned AI tools in access controls (CC5.3), prompt injection (CC5.6), and evaluation of third-party AI models and APIs in vendor management (CC8.1). In Processing Integrity, the points of focus address AI used in system inputs, processing, and outputs.

Points of focus are not individually required, and the draft notes that guidance on applying the criteria to AI components is still being developed. Even so, auditors use points of focus to shape their testing. If AI is part of your product or your internal operations, expect more questions about it in future examinations.

Here is what that is likely to mean for most service organizations:

  • Know where AI is in use. Keep an inventory of the AI features in your product, the AI tools employees use, and the third-party models and APIs your systems call.
  • Control unsanctioned AI use. Define which AI tools are approved, which data can go into them, and how you detect and handle tools that haven't been approved.
  • Treat AI providers as vendors. Assess model providers and AI APIs through the same vendor risk process you use for other critical third parties.
  • Test AI-specific failure modes. If your product uses AI, document how you defend against prompt injection, monitor model behavior, and validate outputs that customers rely on.

If you are already pursuing ISO 42001 or mapping to the NIST AI Risk Management Framework, much of that work will carry over.

Recommended reading

ISO 42001: How to Implement an AIMS for Strong AI Governance

Vendor and third-party risk in the proposed criteria

Vendor management gets its own series in the proposal, and its scope widens to cover the software components and AI services your system depends on.

The proposed CC8.1 adds two points of focus. The first addresses assessing subservice organizations, such as the cloud providers and other service providers your SOC 2 report carves out or includes. The second covers evaluating third-party components, which the draft defines broadly enough to include hardware, software, AI models, APIs, and plugins.

That second point matters because many teams track vendors at the contract level, which can leave component-level dependencies untracked. An open source library, a browser plugin, or a model API may never go through procurement, yet each can affect the security and availability of your system. Under the proposal, auditors have a clearer basis for asking how you identify and evaluate those dependencies.

Two items from the old CC9.1 are handled differently. Business disruption planning moves to Availability (A1.3), and the point of focus on insurance is not carried over.

To get ahead of this, review whether your vendor inventory covers software components and AI services, along with the companies you pay. A software bill of materials (SBOM) and a list of the APIs your product calls are good starting points.

Recommended reading

When the 2027 SOC 2 criteria would take effect

If the AICPA finalizes the criteria in 2027, they would apply to SOC 2 engagements beginning on or after December 31, 2027, with early adoption permitted.

Date Milestone
October 2026 AICPA releases exposure drafts of the proposed 2027 Trust Services Criteria and Description Criteria
January 15, 2027 Public comment period closes
2027 (date not announced) AICPA expected to finalize the criteria, with changes based on comments
December 31, 2027 Proposed effective date for engagements beginning on or after this date

Until then, the 2017 criteria with 2022 points of focus apply. An audit you are scoping now, or a Type II observation period that starts in 2027, will most likely be tested against the current criteria unless you and your auditor agree to adopt early.

The effective date is tied to when an engagement begins, and CPA firms may interpret that differently for Type II reports with observation periods that span the transition. Ask your auditor early which criteria they plan to use for your next report.

The final criteria may also differ from the draft. Exposure drafts often change in response to public comments, so plan for the direction of the proposal and wait for the final text before rewriting your control mappings.

How service organizations can prepare for these SOC 2 changes

The comment period is a good time to understand the draft and find the gaps it would expose, well before any changes take effect. Most of the work falls into a few areas.

Read the draft and decide whether to comment. The exposure drafts are free to download from the AICPA website with an account. If a proposed change would be hard to apply in your environment, the comment period, which closes January 15, 2027, is your chance to say so.

Map your current controls to the proposed structure. Start with the common criteria. Note which of your controls support criteria that are consolidated or removed, and which proposed criteria, such as the expanded change management series, have no clear control behind them yet.

Inventory your AI use. List the AI features in your product, the AI tools employees use, and the models and APIs you depend on. That inventory supports the AI points of focus across risk assessment, access control, vendor management, and processing integrity.

Expand vendor management to components. Make sure your vendor process covers software components, APIs, plugins, and AI models, in addition to service providers under contract.

Review your optional categories. If Availability is in scope, look at how you test resilience and recovery, since the proposal adds a criterion for it (A1.4). If Confidentiality is in scope, confirm you have documented procedures for disposing of confidential data (C1.3).

Talk to your auditor about timing. Ask which criteria your next report will use, whether the firm plans to offer early adoption, and how it will handle a Type II period that overlaps the effective date.

Plan to update your system description. The AICPA also proposed changes to the Description Criteria that govern the system description in your SOC 2 report. Once those details are published, review your description against them.

Organizations that hold other certifications, such as ISO 27001, will also want to watch for an updated AICPA mapping between the Trust Services Criteria and other frameworks once the criteria are final.

What the proposal means for SOC 2 auditors

For CPA firms, the proposal changes the criteria they test against, and it does not change the attestation standards they follow when performing the examination. Auditors will need to update their testing programs, workpaper templates, and report language for the new numbering, criterion titles, and consolidated criteria.

The AI points of focus will likely require the most new work. Firms will need a consistent way to evaluate controls over model monitoring, prompt manipulation, and training data, and the draft notes that guidance on applying the criteria to AI components is still under development.

Auditors will also need to decide how to handle early adoption and engagements that span the effective date, and to communicate that plan to clients well before their next observation period starts.

Recommended reading

Understanding SOC 2 Compliance Requirements

Past SOC 2 updates: the 2022 TSP 100 and DC 200 revisions

The last update before this proposal came in October 2022, when the AICPA revised the points of focus in TSP 100 and the implementation guidance in DC 200. Neither revision changed the criteria themselves.

Date Update What changed
October 2026 Proposed 2027 Trust Services Criteria and Description Criteria (exposure drafts) Criteria consolidated from 61 to 50, common criteria restructured, AI points of focus added
October 2022 Revised points of focus (TSP 100) and revised implementation guidance (DC 200) New and clarified points of focus, mainly for privacy and confidentiality; added examples to description guidance
2017 Trust Services Criteria (TSP 100) Criteria aligned with the COSO 2013 framework; the basis for SOC 2 reports since

The 2022 TSP 100 revisions focused mostly on privacy and confidentiality and clarified how existing criteria apply to newer technologies and risks. For example, the incident response criterion (CC7.4) gained points of focus on applying breach response procedures to confirmed privacy incidents.

The 2022 DC 200 revisions added no new description criteria. They clarified the implementation guidance with more examples, including guidance on when a significant system incident should be disclosed in the system description.

How Secureframe simplifies SOC 2 compliance

While maintaining an updated system of controls is ultimately the responsibility of your own organization, partnering with Secureframe simplifies and streamlines the process.

We save teams hundreds of hours and thousands of dollars spent writing security policies, collecting evidence, hiring security consultants, and performing readiness assessments. And because Secureframe continuously monitors your infrastructure and alerts you of vulnerabilities, you’ll get your SOC 2 report faster and save money while strengthening your security posture. 

Request a demo to learn more about how we can help you get SOC 2 compliant in weeks, not months.

FAQs

Is SOC 2 changing in 2027?

The AICPA has proposed new SOC 2 Trust Services Criteria. They are an exposure draft, not a final standard. If approved, they would apply to engagements beginning on or after December 31, 2027.

What are the current SOC 2 criteria?

The current criteria are the 2017 Trust Services Criteria with revised points of focus from 2022. They remain in effect until the proposed criteria are finalized and take effect.

Will the five Trust Services Criteria categories change?

No. Security, Availability, Processing Integrity, Confidentiality, and Privacy all remain, and Security is still required in every SOC 2 report. The changes are to the individual criteria within those categories.

How many SOC 2 criteria are in the proposal?

The draft has 50 criteria, down from 61. The common criteria drop from 33 to 20, and Availability and Confidentiality each gain one new criterion.

Does the proposal add AI requirements to SOC 2?

It adds AI-related points of focus to existing criteria rather than a new AI criterion or category. The points of focus cover AI governance, model monitoring, prompt manipulation, data poisoning, transparency of AI outputs, training data, and automated decision-making.

Can my organization adopt the new criteria early?

The proposal permits early adoption once the criteria are final. Whether it makes sense for your next report is a decision to make with your auditor.

Do I need to change my SOC 2 controls now?

No. Audits currently in progress and those scheduled before the effective date will use the current criteria unless you adopt early. Use the comment period to review the draft, inventory your AI use, and identify gaps you would need to close.

How do I comment on the proposed criteria?

The AICPA's SOC 2 Working Group is accepting comments through January 15, 2027. The exposure drafts and comment details are on the AICPA & CIMA website.

Anna Fitzgerald

Senior Content Marketing Manager

Anna Fitzgerald is a digital and product marketing professional with nearly a decade of experience delivering high-quality content across highly regulated and technical industries, including healthcare, web development, and cybersecurity compliance. At Secureframe, she specializes in translating complex regulatory frameworks—such as CMMC, FedRAMP, NIST, and SOC 2—into practical resources that help organizations of all sizes and maturity levels meet evolving compliance requirements and improve their overall risk management strategy.

Rob Gutierrez

Senior Cybersecurity and Compliance Manager, CISA, CCSK, CMMC RP

Rob Gutierrez is an information security leader with nearly a decade of experience in GRC, IT audit, cybersecurity, FedRAMP, cloud, and supply chain assessments. As a former auditor and security consultant, Rob performed and managed CMMC, FedRAMP, FISMA, and other security and regulatory audits. At Secureframe, he’s helped hundreds of customers achieve compliance with federal and commercial frameworks, including NIST 800-171, NIST 800-53, FedRAMP, CMMC, SOC 2, and ISO 27001.