REGULATION
FASB and SEC AI Disclosure Rules: What Finance Must Report About AI in 2026

As of mid-2026 there is no dedicated AI disclosure standard from FASB, and the SEC is pointing to its existing framework rather than writing new rules. That does not mean AI is off the disclosure hook. It means the existing materiality and controls rules already apply. This is general information, not legal or accounting advice.
- There is no prescriptive AI disclosure standard in effect. Standard-setters have leaned toward existing GAAP being robust while questioning disclosure sufficiency.
- The SEC has signaled reminders over new rules, pointing to existing materiality, COSO and internal control frameworks.
- AI washing, overstating AI capability or use, is a live comment-letter and enforcement risk.
- Named concerns include third-party embedded AI, data quality and management's own risk assessment, including shadow AI adoption.
- Human review of AI-influenced disclosures is expected. The company retains full disclosure liability.
The question finance teams keep asking is what they now have to disclose about AI. The honest 2026 answer is that no new standard requires anything specific yet, and the existing rules already reach AI where it is material.
That is a more demanding answer than a checklist, because it puts judgment back on the preparer. This is general information and not a substitute for advice from your auditors and counsel.
Where the standard-setters actually landed
Advisory discussions in 2026 leaned toward current GAAP being robust enough, with urgency reserved for questioning whether existing disclosure is sufficient, particularly around large AI infrastructure investment.
The SEC's public signals pointed to reminders rather than prescriptive rules, directing preparers to existing materiality and internal control frameworks rather than a new AI-specific regime.
What that means you should be doing
No new rule does not mean no obligation. The existing framework already covers several things.

AI inside your controls
Where AI influences a number or a control, it falls within internal control over financial reporting. That means the same expectations of documentation, testing and human review that apply to any control.
A model that drafts or suggests is fine. A number that reaches the financials because a model produced it, with no human review and no trail, is a controls problem.
How ChatFin supports the controls story
ChatFin keeps AI inside a governed workflow. Write-back is human approved, the ERP stays the source of truth, and every step carries a prompt to transaction trail, which is exactly the documentation a controls review looks for.
That does not replace your auditors or counsel. It gives them a clean record of what the AI proposed, what a person approved, and why.
No new standard does not mean no obligation. Where AI is material or touches a control, the existing rules already reach it.
Treat AI Within the Rules You Already Have
The 2026 disclosure picture rewards judgment: no new AI standard, but existing materiality, controls and accuracy rules that already apply to AI.
ChatFin keeps AI governed and documented so the controls and disclosure story holds together. Confirm the specifics with your auditors and legal advisors.