Skip to content
Slow Variables

Ask the data

Answers come from the same store as the site; every number is checked against the record it cites.

Adaptation · Deployment & application

FDA-authorised devices built on a large language model

consistent with normal60/95 confidence, mixed or hard to operationalisegrade Aleadingrefreshed irregularly: An event-driven count; it moves only when a new LLM-based device is authorised.

1What this measures

Count of distinct devices the FDA has cleared or authorised whose core function is a large language model, derived metric fda_llm_devices_count over the clearance records. The first is UpDoc (K253281, decision 23 December 2025), per the FDA 510(k) database, STAT News and Innolitics.

Why it matters. The first LLM device through review is the moment the regulated-domain speed limit meets the technology the paper is about; the count after it is the throughput of that limit.

Proxy types
policy
Unit
count
Cadence
quarterly
Valve
adoption to adaptation

2How we track this

  • series fda_510k.updoc.first_llm_device_clearance.pt
  • series statnews.updoc.first_llm_device_clearance.pt
  • series innolitics.updoc.first_llm_device_clearance.pt
  • derived metric fda_llm_devices_count (formula in the semantic layer)
  • source FDA 510(k) premarket notification database · default tier 4 · Public domain (U.S. government work)
  • source Innolitics regulatory articles · default tier 6 · Innolitics; short quotation
  • source STAT News · default tier 5 · STAT; short quotation
Normal band
≤ 3
Fast band
≥ 10
Falsifying

Normal = a handful of LLM-based authorisations by end-2026, given a 510(k) path of roughly six months and a first clearance in December 2025; fast = ten or more, the regulatory throughput a fast scenario needs. Between is `emerging`.

Applied to metric:fda_llm_devices_count.

3Tracker interpretation

One device, cleared on a drug-dose-calculator predicate and publicised six months later. The public record does not itself say "LLM"; Innolitics' caveat stands.

4Evidence

Latest point
1as of 2025-12-23(3 obs)
Value the bands apply to
1as of 2025-12-23(3 obs)

3 observations. Hollow points are disputed (see counterevidence). Every point links to its observation.

Derived rows (1)
as ofdimsvalueinputs
2025-12-231obs:678d971eobs:8da46094obs:f98335f0

5Status and reasoning

consistent with normalsince 2026-09-10 · evaluate

Evaluator: 1 (FDA 510(k) premarket notification database, 2025-12-23) is inside the consistent band (hi=3). Auto-reason; band rationale: Normal = a handful of LLM-based authorisations by end-2026, given a 510(k) path of roughly six months and a first clearance in December 2025; fast = ten or more, the regulatory throughput a fast scenario needs. Between is `emerging`.

6Timeline notes

  • 2025-12-23 · 1as of 2025-12-23(3 obs)

7Counterevidence

What cuts against this reading

The FDA record does not name an LLM; the clearance is narrow (a diabetes dosing app); a count of one is not a rate.

8Update history

  1. 2026-09-10unmeasured to consistent with normalconf 60 · evaluate

    Evaluator: 1 (FDA 510(k) premarket notification database, 2025-12-23) is inside the consistent band (hi=3). Auto-reason; band rationale: Normal = a handful of LLM-based authorisations by end-2026, given a 510(k) path of roughly six months and a first clearance in December 2025; fast = ten or more, the regulatory throughput a fast scenario needs. Between is `emerging`.

9Confidence

60 / 95 — mixed or hard to operationalise

Confidence is independent of status: 90–95 multiple strong independent sources; 70–89 good evidence, some ambiguity; 50–69 mixed or hard to operationalise; below 50 limited or vague.

10Related