Are Next-Gen AI Earbuds a Thyroid Health Risk? The OpenAI × Jony Ive Meets a 2024 Study

In mid-January 2026, a single X post blew up: health coach Ben Smith (@bensmithlive) paired a fresh Weibo supply-chain leak about OpenAI’s rumored “AirPods killer” with headlines from a 2024 Chinese epidemiological paper. The takeaway many readers jumped to? “Always-in-ear AI buds blasting EMF right next to your thyroid = future health disaster.”

The post quickly racked up thousands of likes, hundreds of reposts, and a heated reply thread mixing alarm, skepticism, and wired-earphone evangelism. But how much of this is grounded concern versus early-stage fear-mongering? Let’s break it down.

The 2024 Study That Started the Alarm

Published June 21, 2024 in Scientific Reports (Nature portfolio), the paper titled “Epidemiological exploration of the impact of bluetooth headset usage on thyroid nodules using Shapley additive explanations method” analyzed 600 matched questionnaire responses tied to 96 confirmed thyroid nodule cases.

Researchers used propensity score matching, XGBoost modeling (AUC 0.95), and SHAP value analysis to rank risk factors. Two variables stood out:

• Age (strongest overall predictor, as expected for nodule prevalence)
• Daily Bluetooth headset usage duration — longer use showed a clear positive association with nodule risk

The authors concluded there is a “significant impact relationship” between prolonged use and increased nodules, attributing potential harm to cumulative non-ionizing radiofrequency (RF) exposure from devices worn close to the neck/thyroid region.

Important caveats the paper itself emphasizes:

• Observational design → cannot prove causation
• Self-reported usage data → recall bias possible
• Relatively small effective sample after matching
• No direct dosimetry of RF levels from specific headsets
• No long-term follow-up or biological mechanism proven

Major health agencies (WHO, ICNIRP, FCC) still classify Bluetooth/RF at consumer levels as “possibly carcinogenic” (Group 2B) based on limited evidence — the same category as coffee and pickled vegetables — but routine Bluetooth exposure has not triggered regulatory action.

The Leaked “Advanced Ear” / “Sweetpea” Prototype

The X thread exploded after @zhihuipikachu (a well-followed Weibo consumer-electronics tipster) shared alleged supply-chain details about OpenAI’s first consumer hardware device, developed with ex-Apple design chief Jony Ive after OpenAI’s 2025 acquisition of Ive’s startup “io” for ~$6.5 billion.

Key rumored specs floating around January 12–13, 2026:

• Internal codename: “Sweetpea” (prioritized over earlier “Gumdrop” rumors)
• Form factor: Metal “eggstone/卵石” charging case containing two capsule-shaped “pills” that clip behind the ear (open-ear design in many descriptions, not deep in-canal)
• No classic in-ear stems; audio reportedly uses ultrasonic or bone-conduction-like transmission in some versions
• Target processor: 2nm-class smartphone-grade chip + custom co-processor
• BOM cost approaching flagship phone levels → expected premium pricing
• Ship window: Possibly September 2026, with 40–50 million unit first-year goal
• Part of a broader 5-device roadmap by Q4 2028 (including potential pen and home products)

Early 2025 rumors speculated a deep in-canal “always-on” AI companion, but recent leaks tilt toward behind-the-ear / open-ear architecture — similar to many hearing-aid or sport-audio products — which keeps transducers farther from the thyroid than true in-ear monitors.

Why the Concern Resonates — and Why It’s Premature

The thyroid sits at the base of the neck, roughly 3–5 cm below most earbud speaker drivers. In-ear designs do place antennas and electronics closer than over-ear headphones or phones held at a distance. If the 2024 study’s statistical link holds in future replication (still a big if), any product encouraging multi-hour daily wear could theoretically amplify whatever small effect exists.

Yet several realities temper the panic:

1. Bluetooth Class 1/2 power is extremely low (~1–100 mW peak, far below cell phones).
2. SAR levels for earbuds remain well under safety limits even during prolonged use.
3. No major replication of the Chinese finding has appeared yet (mid-2026).
4. Open-ear designs (the current leak favorite) reduce canal occlusion and move radiators away from neck tissue compared to sealed in-ears.
5. Alternative risks — sustained high volume causes far more proven harm (hearing loss, tinnitus) than theoretical low-level RF.

Bottom Line (for Now)

If you’re already limiting Bluetooth wear time because of the 2024 paper or general EMF caution, the rumored OpenAI/Ive device probably doesn’t change that calculus dramatically — especially if it ultimately ships as an open-ear clip rather than a deep-canal implant.

That said, the combination of always-listening AI + premium hardware + Jony Ive polish could drive dramatically higher average daily usage than today’s AirPods. Should independent RF dosimetry and long-term epidemiological follow-up become standard for next-gen wearables? Absolutely.

Until then: enjoy your music and podcasts, maybe alternate with wired options or speaker mode when possible, and treat single-study associations as interesting signals — not verdicts.

What do you think — game-changing category or solution looking for a problem (and possibly creating new ones)?

(Sources: Scientific Reports 2024; Weibo leaks aggregated via X Jan 12–13 2026; public MacRumors / Verge reporting on Ive/OpenAI project timeline.)