AI Denials in The Newly Proposed NYCEPPO Healthcare Plan
Is our union leadership (the UFT and MLC) ready to handle AI-based healthcare denials? Have they put protections in place for members in this new plan? What proof do we have that they can?
On September 8th, the Municipal Labor Committee gave city union leaders only one day and a handful of hours to examine a redacted version of the newly proposed NYCEPPO healthcare plan.
The documents were thousands of pages long. A lot to digest under duress and leering eyes.
Then two days later, when the MLC met again to discuss the plan in detail, one astute and concerned union leader had a burning question based on what they saw in the plan regarding denials and the use of artificial intelligence (AI). A really good question particularly because UnitedHealthcare is one of the the chosen insurance companies and there are so few existing legal and regulatory oversight protections.
That’s your final answer, Mr. Garrido, Mr. Mulgrew and MLC leadership?
First, there is evidence of an agreement regarding the use of AI in denials in this newly proposed plan.
Second, there is no evidence of discussion, plan or ideas on safeguards concerning the looming implementation of AI in deciding to deny or delay our healthcare?
The MLC will look into it in a committee down the road?
What?
Is the leadership of the MLC prepared for AI denials if they haven’t even thought about it?
This is further proof these leaders are not strategic thinkers. They just want a deal.
And when they are asked good questions they just dismiss them.
They would rather call it fear mongering or appear like they have it all under control.
We don’t know the future, but we want to be confident our leaders are prepared.
Clearly, they aren’t if they are not even thinking about the consequences and impact. While anyone paying even the smallest amount of attention would be concerned about United Healthcare and denials as they are the industry leader in denials and their prolific use of AI to do it.
This isn’t just about some mundane or perfunctory contract clause—it’s about life-or-death decisions being handed to algorithms owned by UnitedHealthcare.
Based on this exchange, it’s in this contract. This is evidence. Not a conspiracy or wild speculation.
Fact: The proposed NYCEPPO plan would allow UnitedHealthcare to use AI to decide whether members get treatment, rationing care by machine.
Delegates must reject this plan unless ironclad safeguards are built in and that we can see exactly what they are.
1. Risk of UHC’s Denied Care
United Healthcare already has a shameful record of putting profits ahead of patients. It leads private insurers with a 32% denial rate. According to an American Medical Association survey, doctors have ranked UHC as the insurer with the most prior authorization hassles, with 72% of physicians giving UHC a “high” or “extremely high” burden rating.
As part of their abysmal record, families have been forced to fight in court after UHC’s denials left them without post-acute care. Recently, lawsuits revealed that UHC’s AI algorithm was denying elderly patients rehab coverage—despite their doctors saying it was necessary. Internal reports showed the company knew the tool was wrong, but kept using it anyway to save money.
Do we really want that same system deciding whether UFT members get care?
2. Opaque Decision-Making & Zero Accountability
When a claim is denied by AI, there’s no one to hold accountable. Members won’t know what data was used, what criteria were applied, or why the decision was made. Even worse—lawsuits have exposed how UHC’s algorithm sometimes overruled medical standards without patients ever knowing. If no one can see inside the black box, how can members ever get a fair fight?
3. Real Harm and Delays in Care
This isn’t theory. UHC’s AI denials have left patients stranded in hospitals, denied rehab, and abandoned in recovery. Doctors have testified in court about patients suffering permanent harm because of delays. Some were discharged too early, fell, and ended up back in the ER. Others never regained full mobility because rehab was cut off mid-stream.
Gary Bent is a man who died in 2023 after his Medicare Advantage plan, administered by UHC’s phased out NaviHealth, repeatedly denied him necessary care following brain surgery for a melanoma recurrence. Gary’s illness, passing and the family’s years long fight against the AI-driven denials led to Megan and her mother, Gloria, testifying before Congress.
Watch this video to hear the Bent family’s tragic story.
This is the future for our members if UHC’s AI is written into this contract.
4. Added Stress and Costs on Members
When UHC’s AI wrongly denies care, the appeals process is brutal. Members are forced to navigate paperwork, hearings, and delays—often while sick or caring for loved ones. Many appeals are eventually overturned, but only after weeks of pain, stress, or financial hardship. By then, the damage is already done.
5. The AMA’s Warning
The American Medical Association has spoken loud and clear: AI must never replace human medical judgment. They’ve called for strict oversight, transparency, and accountability—precisely what this plan lacks. To approve the NYCEPPO as presented is to side with UnitedHealthcare’s cost-cutting algorithms over the voices of doctors and patients.
The Human Cost
For patients and families, these denials aren’t abstract policy disputes. They mean:
A stroke survivor sent home while still needing therapy to walk and talk.
A patient on IV antibiotics discharged before infection was under control.
Families forced to choose between unsafe discharges and draining life savings.
Appeals are technically possible — but they often take weeks or months. Most families don’t have the resources to pay thousands upfront while waiting. By the time an appeal is resolved, the patient’s window for recovery may have closed.
Why This Matters
The promise of AI in healthcare is real: better data, earlier diagnoses, more efficient care. But when algorithms are used to deny services, the incentives flip. Cutting costs can come at the expense of patient safety and well being. If an AI tool consistently underestimates recovery needs, thousands of vulnerable patients could be forced out of care too early.
Urgent Demands
We must demand these protections:
Human Oversight: Every AI denial must be reviewed by a licensed physician.
Transparency: Full disclosure of how the AI makes decisions and what data it uses.
Fast Appeals: Independent reviews with strict timelines.
Monitoring & Reporting: Public numbers on denials, overturned cases, and patient harm.
No Black Box Models: No secret or unaudited algorithms allowed.
Medical Standards First: Care decisions must be guided by doctors, not corporate cost formulas.
Call to Action for UFT Delegates
Vote NO unless UHC’s AI clauses are stripped out or heavily amended with strong protections.
Vote NO because this all points to one unavoidable conclusion: this plan cannot be trusted — including, their refusal to show all of us the unredacted contracts, the many risks of a reckless self-funded gamble, and Mulgrew’s record of betrayal with his failed Medicare Advantage plan.
Educate Members: Show them what AI-driven denials have already done to patients across the country.
Demand Accountability: Hold leadership responsible for protecting members, not insurers.
Stand with Physicians: Join the AMA in fighting back against insurer abuse of preauthorizations, including the use of AI.
Bottom Line: Protect Care, Not Algorithms
United Healthcare has already been sued for using AI to wrongly deny care. Judges, patients, and doctors have all sounded the alarm: this system harms people. Now, NYCEPPO wants to hand them the keys to our healthcare to UHC’s digital algorithms.
Delegates must stand firm. This is about defending our right to care determined by doctors—not by a cost-cutting computer.
Speak to your CL and your delegates about delaying our Delegate Assembly vote on Monday, September 29th.
We deserve better than this.
Vote NO. Protect our members. Protect our families. Protect patients. Protect our dignity.
References: UnitedHealthcare’s Record on AI Denials
Class-Action Lawsuit Filed Against UnitedHealth & NaviHealth -
Allegations that UHC’s nH Predict AI tool systematically denied post-acute care despite doctors’ orders; ~90% of denials were later overturned.
Source: STAT News, Nov. 14, 2023Federal Court Allows Case to Proceed -
Minnesota judge ruled breach-of-contract and good-faith claims against UHC over AI denials may go forward.
Source: Digital Healthcare Law, Mar. 27, 2025Estate of Gene B. Lokken v. UnitedHealth -
Lokken, age 91, denied needed rehab coverage; family alleges harm and financial devastation caused by UHC’s algorithm-driven decisions.
Source: Financial Times, Nov. 2023Senate Investigation: Sharp Spike in Denials Under NaviHealth -
Denial rate for skilled nursing facility admissions jumped from ~1.4% in 2019 to ~12.6% in 2022 after NaviHealth implemented algorithm-based controls.
Source: StatNews, March 17, 2024Patients Harmed by AI-Driven Denials -
Families describe devastating consequences: rehab cut off too soon, patients discharged prematurely, long-term recovery lost.
Source: STAT News, May 19, 2025Court Denies UHC’s Bid to Limit Discovery -
Judge ruled UHC must turn over internal documents on how its AI tools make decisions.
Source: Becker’s Payer, Apr. 2025
The beginning of the campaign to win the next union election and build support for using a threat of a teacher strike to frighten the city to roll over.
The “vote down the plan” means an angry lame duck mayor, Adams might be able to impose a plan as his final “gift?”
Attacking the union from within only strengthens our enemies