FDA's CDRH Flags AI and Surgical Robots for FY2027

FDA's device center wants three draft guidance documents finalized before anything else next year, and has opened a file for a kind of device it hasn't cleared a single one of yet. CDRH published its proposed Fiscal Year 2027 guidance agenda on 1 October 2026, the annual list that tells device sponsors which of the agency's draft positions are closest to becoming final policy — and where its regulatory thinking is headed next.

Why a guidance agenda is worth reading closely

A guidance agenda is not a rule and binds nobody, but it is the clearest public signal FDA's device center gives of which draft positions it considers close to settled and which open questions it is prioritizing. For a sponsor building a submission strategy around AI-enabled software, a change-control plan, or a surgical robot, the agenda is the difference between planning around a position that could remain a draft indefinitely and one FDA has told the public it intends to finalize within the year.

The three at the top

• AI-enabled device software functions: finalizing total-product-lifecycle management guidance first drafted in January 2025 — how a device's AI model is expected to be monitored, updated and revalidated across its life, not just at clearance.
• Predetermined change control plans: finalizing guidance first drafted in August 2024 on what a sponsor must pre-specify to make certain post-market software or algorithm changes without a new submission.
• Robotic-assisted surgical devices: finalizing guidance covering submission content for RASDs, the category that includes the draft FDA separately issued in late September 2026 on instrument control, latency, cybersecurity and sterilization.

A new file before a single application exists

The agenda's newest addition has no finalized predicate to build on: draft guidance on evidentiary recommendations for generative-AI-enabled conversational devices intended for mental health disorders. FDA has authorized more than 1,200 AI-enabled medical devices overall, and none yet use generative AI for a mental-health indication. Writing guidance ahead of a first application, rather than reacting to one, is itself a signal — CDRH expects sponsors in this space soon enough to want the evidentiary bar defined before the first submission arrives, not negotiated case by case.

What sponsors should do with it

The agenda carries no legal force, and FDA regularly misses its own guidance timelines. Treat it as a planning input: a sponsor relying on an AI/ML lifecycle approach, a change-control plan, or a robotic-surgery submission now has a declared one-year window to watch for a final document, and a generative-AI mental-health developer has an unusually early warning of what FDA will expect to see. None of that changes today's review standards; it changes which drafts are worth building a submission strategy around before they are final.

Frequently asked questions

What did CDRH release on 1 October 2026?

Its proposed guidance agenda for Fiscal Year 2027 (October 2026–September 2027), listing the device guidance documents it intends to finalize or advance over the coming year. Public comment is open through 30 November 2026.

What are the top priorities?

Finalizing guidance on AI-enabled device software total-product-lifecycle management, on predetermined change control plans, and on robotic-assisted surgical devices — all already circulated as drafts.

What is new on the list?

Draft guidance on evidentiary recommendations for generative-AI-enabled conversational devices for mental health disorders — written ahead of any application in this category, since none has been authorized yet.

Does this duplicate FDA's robotic-surgery draft guidance from late September?

No. That was a separate document on RASD submission content. The FY2027 agenda lists finalizing it as a priority, a later step, not a re-announcement.