Oopsie #1 Philips Respironics: The Foam That Poisoned Millions of Sleepers

- What happened: Philips built sound-dampening foam into its CPAP, BiPAP, and ventilator machines to reduce noise. A reasonable idea. Except the foam, a polyester-based polyurethane (PE-PUR), degraded over time, especially in warm, humid environments (you know, like the inside of a breathing machine that sits in your bedroom). The foam broke down into particles and toxic gases, including formaldehyde, which users were then inhaling while they slept. Every night. For years.
- The scale: The recall covered roughly 15 million devices sold between 2008 and 2021. FDA received over 116,000 Medical Device Reports, including 561 deaths, connected to the foam degradation. Philips knew there were complaints as early as 2015 and didn't issue the recall until June 2021.
- The aftermath: Philips agreed to a $1.1 billion personal injury settlement, a separate $479 million class-action settlement for economic losses, and a consent decree with FDA and the DOJ that banned them from selling new CPAP and BiPAP devices in the US until conditions were met. They stopped selling entirely. Philips, one of the most trusted names in healthcare, effectively exited the US sleep apnea market they had dominated for years.

Oopsie #2 Life-Saving Devices Are Also The Most-Recalled Devices.

- What happened: A 2024–2025 analysis of FDA recall data found that the three most frequently recalled device categories were ventilators, resuscitators, and infusion pumps: all Class I recalls, meaning the most serious type, with potential for serious injury or death. Philips topped the manufacturer list with 5 recalls affecting 383,141 units. Smiths Medical came in second with 4 recalls affecting nearly 1 million units.
- Why this matters: These aren't fringe products. Ventilators keep people alive when they can't breathe. Infusion pumps deliver precise medication doses. Resuscitators restart hearts. The fact that these categories dominate the recall list isn't a coincidence, they're complex, used in high-stakes environments, and often aging in hospital storage rooms for years between maintenance cycles.

Oopsie #3 STRYKER - When A Medical Device Gets Wiped Out by Hackers

- What happened: On March 11, 2026, an Iran-linked hacking group called Handala launched a cyberattack on Stryker, a Fortune 500 medical device maker that makes everything from orthopedic implants to defibrillators. The attackers claimed to have wiped more than 200,000 devices, servers, and laptops across 79 countries. They used Stryker's own Microsoft Intune device management system to remotely wipe thousands of employee devices. The attack claimed to be retaliation for US military action in Iran.
- The impact: Order processing, manufacturing, and shipping ground to a halt. Surgeries were rescheduled because custom implant orders couldn't be fulfilled. Stryker's LIFENET system, used by emergency responders to transmit ECG data to hospitals, went temporarily offline in parts of Maryland. CISA issued a national alert urging healthcare organizations to harden their endpoint security in direct response to the Stryker attack.
- What Stryker said: The devices themselves were safe. No ransomware. The attack was contained to internal IT systems. Stryker filed an 8-K with the SEC and brought in Palo Alto Networks Unit 42 to investigate. Estimated device restoration cost alone: $24–40 million, before you count lost productivity, incident response, or delayed surgeries.

Oopsie #4 AI Medical Devices: Cleared Fast, Recalled Faster

- What happened: A Johns Hopkins study examined 950 AI-enabled medical devices authorized by FDA through November 2024. Of those, 60 devices were associated with 182 recall events. The most common causes: diagnostic errors and functionality failures. The most alarming finding: 43% of all recalls happened within one year of FDA clearance. The majority of recalled AI devices had not gone through clinical trials before reaching the market.
- Why: Most AI devices reach the market through the 510(k) pathway, which doesn't require clinical trials. If the algorithm is deemed "substantially equivalent" to a predicate, it can get cleared without prospective human testing. That works fine for a lot of devices. For AI tools making diagnostic decisions? The gap between lab performance and real-world performance can be brutal.
- Publicly traded companies accounted for 53% of recalled AI devices and 98.7% of recalled units. Researcher Tinglong Dai from Johns Hopkins attributed this to investor-driven pressure for faster launches.