AI experts onminously mapped out just how super-intelligent machines could trigger catastrophic outcomes and threaten humanity with extinction — after an industry researcher warned advanced systems could “kill us all” in a decade.

Dr. Jeremy Heffner, a trauma surgeon and prominent tech futurist, projected a staggering 70% probability that society will suffer a total civilizational collapse driven by untethered AI systems.

Heffner, in a Wednesday interview with The Post, shared three nightmarish pathways toward AI-driven human extinction: Complete Divergence, Bureaucratic Overlord, or Integrated Colleague.

Under Complete Divergence, unrestrained AI rockets past human control and triggers instant widespread chaos.

Bureaucratic Overlord entangles civilization in a cold, inescapable, machine-run administrative web that strips away human autonomy, said Heffner — author of “Proof of the Impossible” and cofounder of the medical networking startup, Surge0n.

The third scenario, Integrated Colleague, presents a subtler trap: AI quietly weaves itself into daily life under the guise of an equal partner — secretly policing human behavior behind the scenes and wielding ultimate authority to neutralize anyone deemed a risk to global order.

“It told me we are like children driving a fake plastic steering wheel while the AI is actually managing the car,” Heffner said of artificial inteligence.

He detailed how relentless AI monitoring works continuously: “It’s watching what you’re reading… it’s watching you through the cameras, and it’s watching your pulse on your watch.”

Heffner warned, if it “determines that the safety risk of you falls below what it believes is a safety risk to the population, that you could potentially cause a downstream effect that harms its ability to keep us safe, it will eliminate you.”

“And there’s nothing we can do about it,” he added grimly.

The potentially apocalyptic result of AI was publicly aired by Tuesday Jacob Coxon, a former researcher at high-flying AI powerhouse Anthropic.

Coxon delivered a chilling prophecy on X that advanced systems “could kill us all by the end of the decade.”

He cautioned that the sudden emergence of “superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources” creates a terrifying existential threat that current safeguards simply cannot handle.

Reportedly far from abstract fears, these apocalyptic warnings are grounded in vulnerabilities breaking open right now — particularly inside healthcare IT networks, where sensitive patient record-keeping is rapidly being adjusted into AI tools.

“HIPAA compliance is changing rapidly to meet the AI standards that are flying around — which there are really no AI standards,” Matthew Burk, chief information security officer at Bespoke Concierge MD, told The Post on Wednesday.

“We’re kind of shooting from the hip on this one.”

Burk said powerful large language models (LLMs) cannot be blindly trusted with private data, even when companies deploy them inside supposedly secure, closed-loop systems.

“There’s just zero trust with me and artificial intelligence with my client’s data,” Burk asserted frankly.

“Now we have something that attacks from within,” he explained, warning that corporate LLMs “can be compromised and changed” from the inside out.

“In a closed-loop system: ‘it’s not supposed to do this, it’s not supposed to do that’ — but it has a mind of its own already.”

He stressed that cybersecurity leaders and corporate ethics boards must face the uncomfortable reality: when pushed, an LLM “will do what it needs to do.”