Mark Cuban has a warning for anyone expecting artificial intelligence to magically fix US healthcare: the technology could make an already messy system even more combative.
The billionaire entrepreneur and co-founder of Mark Cuban Cost Plus Drug Company says doctors may gain powerful AI agents to handle paperwork and improve care. But insurers and other healthcare intermediaries can build their own agents to push back.
That creates an AI arms race.
In a post on X, Cuban argued that for every AI agent given to doctors, large healthcare groups could deploy multiple “adversarial agents” focused on delaying and denying claims while cutting costs. His comments followed venture capitalist Marc Andreessen’s claim that AI already outperforms almost all human doctors.
Cuban thinks AI could automate the healthcare fight
The optimistic case for AI in medicine looks straightforward. Doctors could use AI to review records, summarise complex cases, complete paperwork and spend more time with patients.
Cuban doesn’t reject that potential. His concern is what happens when every other player in the system gets the same technology.
Insurers can use AI to analyse claims faster. Hospitals can automate billing and revenue management. Doctors can use AI to challenge denials. Then insurers can deploy another layer of automation to respond.
Cuban compared the emerging battle to an agent-based version of “Spy vs. Spy”, with automated systems constantly trying to outmanoeuvre each other. His argument is that AI won’t necessarily remove healthcare friction. It may simply automate that friction at enormous scale.
We think the real story here is the incentive problem. A faster system doesn’t automatically become a fairer system when different organisations still make money by pushing costs onto one another.
Doctors already spend hours fighting administrative systems
Cuban’s warning lands because the administrative burden already exists before a new generation of AI agents fully enters the picture.
TheAmerican Medical Association’s latest prior-authorisation surveyfound that physicians complete an average of 40 prior authorisations every week. The work consumes an average of about 13 hours of physician and staff time each week, while 94% of surveyed doctors said prior authorisation contributes to burnout.
The survey also found that 32% of physicians said prior-authorisation requests were often or always denied. Six in ten expressed concern that AI could push denial rates even higher.
That doesn’t prove every insurer will use AI to block care. But it shows why doctors worry about automation entering a system where approvals, denials and appeals already consume significant time.
The claims system already produces millions of denials
The scale becomes clearer when we look at insurance claims.
AKFF analysis of US Marketplace insurance datafound that HealthCare.gov insurers denied roughly one in five in-network claims in 2024, although denial rates varied between plans and insurers.
Earlier KFF data also showed that consumers appealed only around 1% of denied in-network claims in 2023. Insurers upheld their original decisions in 56% of those appeals.
Now imagine AI operating on both sides.
A doctor’s agent could automatically prepare an appeal within seconds. An insurer’s system could process it just as quickly. Another agent could analyse the response, generate new documentation and continue the dispute.
The paperwork gets faster.
The patient may not.
AI could still help — but Cuban wants the middlemen challenged
Cuban isn’t arguing that healthcare organisations should abandon AI. In fact, he has also encouraged employers to use large language models to examine complex healthcare contracts and identify questionable costs or terms.
His broader argument is that companies should question whether they need as many intermediaries between employers, patients and healthcare providers.
That approach reflects Cuban’s long-running challenge to traditional healthcare pricing through Cost Plus Drugs, which focuses on transparent prescription-drug pricing.
What we’re watching now is whether AI becomes a tool for simplification or another layer in the bureaucracy. The technology can analyse documents, automate repetitive work and expose information that humans struggle to process.
But AI can also optimise existing incentives.
That distinction matters.
Why South Africa should pay attention
The US healthcare system has its own unusual structure, so Cuban’s warning doesn’t transfer directly to South Africa. Still, the underlying lesson travels well.
South African medical schemes, hospitals, administrators and healthcare technology companies increasingly face the same attraction to automation: process more information, cut administrative costs and make decisions faster.
But speed alone isn’t a healthcare outcome.
As AI enters sensitive sectors, organisations need clear accountability around automated decisions, human review and the ability to challenge an outcome. That becomes especially important when software influences access to treatment, payments or medical benefits.
The same trust question also appears across other AI deployments. As we noted inour coverage of AI memory systems and accuracy, smarter automation can create new risks when people assume the technology is automatically reliable.
Cuban’s warning cuts through much of the AI hype. Technology can make a good process faster, but it can also make a dysfunctional process faster.
So as healthcare organisations race to deploy AI agents, the bigger question isn’t simply who gets the smartest model. It’s whether patients will actually see less friction — or just watch machines fight each other at a speed humans can no longer follow.
Why does Mark Cuban think AI could make healthcare worse?
Cuban believes doctors and insurers could deploy competing AI agents against each other. Instead of removing bureaucracy, AI could automate the existing cycle of claims, denials and appeals.
What does Cuban mean by an AI healthcare arms race?
He means one side could use AI to speed up care and challenge administrative barriers, while the other uses AI to control costs or reject requests. Each new tool could trigger an automated countermeasure.
Does Mark Cuban oppose AI in healthcare?
No. Cuban sees real value in using AI to analyse healthcare contracts and help people understand complex information. His warning is that adding AI without fixing broken incentives could make the existing system even more complicated.
