Fierce Biotech: AI model designs new treatment candidate for opioid addiction that cuts cravings in rats

Fierce Biotech: AI model designs new treatment candidate for opioid addiction that cuts cravings in rats

By Darren Incorvaia  Apr 6, 2026 3:00pm

While the opioid epidemic is one of the most urgent public health emergencies facing the U.S., new therapeutic approaches for treating the underlying substance use disorder have lagged far behind. Scientists have now used an artificial intelligence program to design a compound that reduced fentanyl use in addicted rats, which could one day serve as a new tool to help patients recover from opioid abuse.

California-based company GATC Health used an AI platform called Operon to scan through data from the brains of people who had lived with opioid use disorder for potential drug targets. The model homed in on two serotonin receptors and then repeatedly crafted better and better molecules that can bind to them, according to a new research paper.

A team led by Christie Fowler, Ph.D., a neuroscientist at the University of California, Irvine, then tested the ability of the two most promising candidates to reduce opioid use in rats addicted to the drug. One compound in particular, GATC-1021, was able to cut cravings for opioids in the rodents with no noticeable behavioral or physical side effects.

The results were published in the Proceedings of the National Academy of Sciences.

Fowler was skeptical of the AI model at first, she told Fierce Biotech, “as I think any good scientist is.”

“But once we started testing the compounds in our lab,” she added, “we were pleasantly surprised that they worked.”

She also heralded the model’s potential to speed up the famously lethargic drug development process. Rather than starting with 50 compounds to test, as she’s used to, her team only had to work with two.

GATC-1021 targets two serotonin receptors, 5HT2A and 5HT6, which, when targeted alone can produce hallucinogenic effects. But Fowler’s team noticed no such psychedelic effects of GATC-1021, likely because the AI model had fine-tuned the degree to which the compound engaged each receptor.

Currently, the standard drugs used to address opioid use disorder are themselves opioids, which adds to the stigma around addiction and makes accessing treatment challenging. The most common opioid used in this harm reduction approach is methadone, which has existed since the 1960s.

“What if you had to go somewhere, wait in a line, stand there while you’re dosed, and do that every single morning of your life?” Stephen Loyd, M.D., an internal medicine and addiction medicine physician at Cedar Recovery in Tennessee, told Fierce. “If you had a tool that didn’t carry that baggage, it’s not only a needle changer, it is a paradigm shifter.”

In addition to not being an opioid, Fowler’s team also found that GATC-1021 promoted the growth of new connections between neurons in the rats’ brains in regions associated with learning and memory. This neuronal refreshing in the brain may help break the entrenched behavioral patterns that fuel addiction.

“Drug use is a really complicated thing that begins as this high level of reinforcing behavior, but then you get a lot of learned associations that’s really propagating relapse,” Fowler explained. For example, someone in recovery from alcohol use disorder may develop a craving to drink after walking by a bar that they used to frequent.

By altering the brain’s architecture itself, GATC-1021 could not only stave off cravings but also help patients break these behavioral cycles that often trigger relapse. Loyd called this potential a “quantum leap” in addiction treatment.

“The longer you can keep somebody in the recovery process, the better their long-term outcomes,” said Loyd, who has himself been in recovery from opioid use for more than 20 years after becoming addicted to OxyContin and Xanax during the final year of his internal medicine residency.

“What makes GATC-1021 so compelling is that it was designed from human data to target the specific neurological drivers of opioid dependence, rather than simply substituting one opioid for another,” Rahul Gupta, M.D., president of GATC Health and former director of the White House Office of National Drug Control Policy, said in a press release shared with Fierce. “That level of precision is only possible through AI, and it represents a fundamentally new way to think about treating this disease.”

Fowler’s team put together a robust preclinical package for GATC-1021, and the plan now is to move forward with an investigational new drug application and clinical trials, she told Fierce. Fowler hopes to also test the compound’s potential in other addictions and psychiatric diseases, too, given the similar mechanism underlying the conditions.

To Loyd, who was not involved with the work but had previously met GATC Health’s leadership at a conference in Washington, D.C., the company is doing important work in a patient population that has long been neglected by the pharmaceutical industry.

“Industry has shown no interest in this field other than pain medication,” he said, noting stigma against addiction and the difficulty of reaching patients as key reasons why. But the potential impact of a new opioid addiction treatment is huge, as the disease does not discriminate.

“You’re talking about millions and millions of lives,” Loyd said. “I don’t think they understand the scope of the market.”