46 articles · Data & AI · page 1 of 3
Articles
Week of 14 September 2026
Hospital device data rights arrive one purchase order at a time
From September 12, 2026, the EU Data Act required new connected products to give users their data by design, but only for equipment marketed from then. Hospitals replace scanners slowly, so tenders carry the new right while older equipment runs on requests.
Week of 13 July 2026
Fedwire’s ISO 20022 switch gave banks rich payment data their core systems still truncate
A year after Fedwire switched to ISO 20022, the richer payment message reached most US banks intact. The value inside it often did not, lost at the customer’s keyboard, inside banks’ own systems, or in screening and fraud tools never set up to read it.
Week of 6 July 2026
Medicaid’s frailty test asks for a code and proof of impairment, and clinicians hold both
CMS’s June 2026 rule tied Medicaid’s “medically frail” exemption from work requirements to a coded diagnosis from the past year plus evidence the condition limits work. Suddenly, medical coding, doctors’ paperwork and data sharing help decide who keeps their coverage.
Week of 23 March 2026
The tool that checked for supply-chain risk was the supply-chain risk
Hackers turned a trusted security scanner into a password thief, then used what it stole to poison a popular AI gateway. Exposure came down to three things: which tools held the keys, where keys were pooled, and how long stolen keys stayed valid.
Week of 5 January 2026
FDA’s decision support revision handed hospitals a classification question
Headlines called FDA’s January 2026 revisions deregulation. The reality was subtler: the agency chose not to police more decision support software, which left hospitals, not regulators, deciding which category each tool falls in, and checking that it stays there.
Week of 8 December 2025
Why the AI preemption order barely reached health care’s own AI rules
December 2025’s executive order aimed federal pressure at broad state AI laws. But most AI rules health systems actually follow sit in licensing, insurance and privacy law, where preemption arguments are weakest. The order’s real risk ran through vendors.
Week of 17 November 2025
For EU consumer lenders, the credit-scoring deadline that mattered was never the AI Act’s
Brussels proposed delaying the AI Act’s high-risk rules, and credit scoring seemed to get breathing room. For consumer lenders it was a mirage: the GDPR, as Europe’s top court read it, and the Consumer Credit Directive’s November 2026 date still set the pace.
Week of 15 September 2025
Approval is easy, monitoring is hard: the Joint Commission’s AI guidance and the work after go-live
In September 2025, the Joint Commission and the Coalition for Health AI issued guidance that looked like a governance checklist. Most of its weight fell after go-live, on monitoring and safety reporting that depend on vendor contract terms few hospitals had negotiated.
Week of 11 August 2025
Secrets in a public docket: the court filing breach and where the sensitive tier ends
The federal courts already had a separate path for their most sensitive documents when their filing system was breached in 2025. The trouble was where its boundary sat. Agencies holding dangerous records in general systems need a tier defined by consequence, not label.
Week of 2 June 2025
Once FDA read with AI, sponsors’ inconsistencies got cheaper to find
FDA gave its reviewers Elsa, an AI assistant for searching and comparing documents, and the inconsistencies in drug submissions became cheaper to find. Sponsors need consistent dossiers, a fast way to correct misreadings, and proof their own AI works.
Week of 31 March 2025
Lighter rules, heavier contracts: OMB’s April 2025 AI memos handed governance to the buyer
OMB’s April 2025 memos lightened internal AI process but asked contracts to carry more of the weight. Agencies that treated every stage of a contract as a governance tool kept bargaining power that review boards alone could not give them.
Week of 17 March 2025
23andMe’s bankruptcy turned a privacy policy into a term of sale
Once 23andMe filed for Chapter 11, its privacy promises stopped being fine print and became a condition that courts, states and bidders had to put a price on. Anyone holding or sharing health-related data should write contracts with that ending in mind.
Week of 20 January 2025
Revoking the AI executive order could not revoke the AI inventory, because Congress had written it
On inauguration day the 2023 AI executive order was revoked, and replacement guidance was weeks away. Agency AI programs could sort their work by what held it up: the order alone, OMB memos awaiting revision, or laws Congress had passed that still applied.
Week of 6 January 2025
Under FDA’s change-control model, the AI device a hospital validates will not stay the one it runs
FDA’s January 2025 draft on AI devices, read with its December change-control guidance, assumes the software keeps changing after a hospital buys it. Three kinds of change follow: pre-approved updates, updates needing fresh review, and shifts in the hospital itself.
Week of 2 December 2024
The CFPB’s data broker proposal exposed a provenance gap that AI programs inherit
A federal plan to regulate many data brokers like credit bureaus was withdrawn within six months. The question it asked, where did this personal data come from and what may we do with it, now follows every AI program.
Week of 9 September 2024
Why o1’s reasoning summaries added work for clinical AI committees
OpenAI’s o1 explained its reasoning just as federal rules on clinical algorithm transparency were taking hold. A fluent explanation answers the clinician’s question, not the vendor’s or the regulator’s, so health systems still had to supply two other kinds of transparency themselves.
Week of 20 May 2024
How Windows Recall put a possible public record where no records officer could reach it
Microsoft answered Recall’s security critics by keeping screen snapshots on the laptop, locked to the user and set to expire. Those same three choices put the snapshots beyond the search, hold and retention tools agencies rely on for public records and lawsuits.
Week of 25 March 2024
Easy to grant, hard to defend: M-24-10 put every federal AI exception on the record
OMB’s final AI memo kept the hard deadline: make risky AI safe or switch it off. It also made every way out, whether a determination, a waiver or an extension, carry a named signer, written evidence and a record others would read.
Week of 11 March 2024
Why the AI Act left bank credit models with the supervisor they already had
Europe’s new AI law did not hand bank credit models to a new regulator. It routed them through the banking supervisor banks already knew, and let existing governance count for some duties. The real work was finding what model risk management never covered.
Week of 11 December 2023
For hospitals, HTI-1’s model cards covered only the EHR vendor’s own algorithms
In December 2023, a federal health IT rule made electronic record vendors explain the predictive tools they supply. For models a hospital built itself or bought elsewhere, the record system only had to hold the information, and the hospital had to produce it.