OpenAI is partnering with Los Alamos National Laboratory to study how artificial intelligence can be used to fight against biological threats that could be created by non-experts using AI tools, according to announcements Wednesday by both organizations. The Los Alamos lab, first established in New Mexico during World War II to develop the atomic bomb, called the effort a “first of its kind” study on AI biosecurity and the ways that AI can be used in a lab setting.

The difference between the two statements released Wednesday by OpenAI and the Los Alamos lab is pretty striking. OpenAI’s statement tries to paint the partnership as simply a study on how AI “can be used safely by scientists in laboratory settings to advance bioscientific research.” And yet the Los Alamos lab puts much more emphasis on the fact that previous research “found that ChatGPT-4 provided a mild uplift in providing information that could lead to the creation of biological threats.”

Much of the public discussion around threats posed by AI has centered around the creation of a self-aware entity that could conceivably develop a mind of its own and harm humanity in some way. Some worry that achieving AGI—advanced general intelligence, where the AI can perform advanced reasoning and logic rather than acting as a fancy auto-complete word generator—may lead to a Skynet-style situation. And while many AI boosters like Elon Musk and OpenAI CEO Sam Altman have leaned into this characterization, it appears the more urgent threat to address is making sure people don’t use tools like ChatGPT to create bioweapons.

“AI-enabled biological threats could pose a significant risk, but existing work has not assessed how multimodal, frontier models could lower the barrier of entry for non-experts to create a biological threat,” Los Alamos lab said in a statement published on its website.

The different positioning of messages from the two organizations likely comes down to the fact that OpenAI could be uncomfortable with acknowledging the national security implications of highlighting that its product could be used by terrorists. To put an even finer point on it, the Los Alamos statement uses the terms “threat” or “threats” five times, while the OpenAI statement uses it just once.

“The potential upside to growing AI capabilities is endless,” Erick LeBrun, a research scientist at Los Alamos, said in a statement Wednesday. “However, measuring and understanding any potential dangers or misuse of advanced AI related to biological threats remain largely unexplored. This work with OpenAI is an important step towards establishing a framework for evaluating current and future models, ensuring the responsible development and deployment of AI technologies.”

Reached for comment over email, a spokesperson for OpenAI tried to emphasize the idea that artificial intelligence itself isn’t a threat, suggesting that misuse of AI was the real threat.

“AI technology is exciting because it has become a powerful engine of discovery and progress in science and technology,” the OpenAI spokesperson said. “While this will largely lead to positive benefits to society, it is conceivable that the same models in the hands of a bad actor might use it to synthesize information leading to the possibility of a ‘how-to-guide’ for biological threats. It is important to consider that the AI itself is not a threat, rather it is how it can be misused that is the threat.”

This idea that AI itself isn’t a threat is, of course, at odds with what Altman himself has said in the past.

“Previous evaluations have mostly focused on understanding whether such AI technologies could provide accurate ‘how-to-guides’,” the spokesperson continued. “However, while a bad actor may have access to an accurate guide to do something nefarious, it does not mean that they will be able to. For example, you may know you need to maintain sterility while cultivating cells or use a mass spec but if you do not have experience in doing this before, it may be very difficult to accomplish.”

And that’s where the statement from OpenAI’s spokesperson really tried to pivot back to the original message that this is all about better understanding lab work.

“Zooming out, we are more broadly trying to understand where and how [do] these AI technologies add value to a workflow,” the spokesperson said. “Information access (e.g., generating an accurate protocol) is one area where it can but it is less clear how well these AI technologies can help you learn how to do a protocol in a lab successfully (or other real-world activities such as kicking a soccer ball or painting a picture). Our first pilot technology evaluation will look to understand how AI enables individuals to learn how to do protocols in the real world which will give us a better understanding of not only how it can help enable science but also whether it would enable a bad actor to execute a nefarious activity in the lab.”

Only time will tell whether this idea holds water that you shouldn’t blame AI, but rather the people who misuse AI. It’s a reasonable position for most technological advances right up until you consider the case of nuclear weapons.

 

A popular class of diabetes and obesity drugs may help keep certain cancers at bay, new research suggests. Scientists have found evidence that GLP-1 medications are associated with a lower risk of 10 different obesity-related cancers in people with type 2 diabetes, compared to only taking insulin. Newer, more potent GLP-1 drugs such as semaglutide and tirzepatide could provide an even greater buffer against cancer, the scientists say.

Glucagon-like peptide-1 receptor agonists (GLP-1RAs) have been used since the early 2000s to treat type 2 diabetes. They mimic the naturally occurring hormone GLP-1, which helps regulate our metabolism by prompting the release of insulin when needed. As it turns out, the hormone also plays a role in regulating our appetite, leading to the approval of GLP-1 drugs as treatments for obesity (the first official medication in the U.S. was liraglutide in 2014). More recent drugs such as semaglutide (the active ingredient in Ozempic and Wegovy) and tirzepatide (Mounjaro/Zepbound) have proven to be substantially more effective at helping people lose weight than diet and exercise alone. In clinical trials, people using these drugs have lost an average of 15% to 20% of their body weight over a year’s time.

This new study was led by researchers from Case Western Reserve University in Ohio. They were interested in studying the health benefits of GLP-1 drugs beyond diabetes and obesity. Because obesity has been tied to a greater risk of 13 cancers, including breast, thyroid, and pancreatic cancer, the team wanted to know if the medications would provide a preventive anti-cancer effect as well. Their earlier work had suggested that GLP-1s are associated with a lower risk of developing colorectal cancer specifically. But as far as they know, this is the first research to systematically look at the link between obesity-related cancers and GLP-1 use.

The scientists analyzed medical record data from 1.6 million patients diagnosed with type 2 diabetes and no pre-existing cancer history. The team studied three subsets of patients who were prescribed different drugs to manage their diabetes sometime between 2005 to 2018: people who were prescribed a GLP-1 drug, people prescribed insulin, or people prescribed metformin, another common diabetes medication. The latter two groups acted as a sort of control comparison.

Overall, the team found that people prescribed GLP-1s had a noticeably lower risk of developing 10 out of the 13 cancers associated with cancer, including gallbladder, pancreatic, liver, and colorectal cancers, when compared to those taking insulin. GLP-1 users may also have had a lower risk of several cancers when compared to metformin users, though the differences were not statistically significant. The team’s findings were published Friday in the journal JAMA Network Open.

This sort of research can only demonstrate a correlation between GLP-1 use and cancer risk, not prove a cause-and-effect relationship. The study can’t also tell us why GLP-1s might protect us from cancer, though the researchers already have some possible explanations in mind.

“The protective effects of GLP-1 RAs against obesity-associated cancers likely stem from multiple mechanisms,” lead author Lindsay Wang, an undergraduate in Case Western Reserve University’s Pre-Professional Scholars Program (a 8-year-long program that combines a Bachelor’s and medical degree), told Gizmodo in an email. “Primarily, these drugs promote substantial weight loss, reducing obesity-related cancer risks. They also enhance insulin sensitivity and lower insulin levels, decreasing cancer cell growth signals.”

Not all the team’s findings were good news. Compared to those on metformin, people on GLP-1s did appear to have a higher risk of kidney cancer. There was also no difference in the risk of thyroid cancer between insulin and GLP-1 users. But since some evidence has found that high insulin levels in the body can raise thyroid cancer risk, that might suggest GLP-1 use can do the same. Notably, some studies in animals have found that GLP-1 drugs can increase the odds of a specific, rarer type of thyroid cancer: medullary thyroid cancer. But the evidence of this added risk in humans has been inconclusive so far.

Even if these negative cancer risks are causally linked to GLP-1 use, though, their anti-cancer benefits elsewhere should likely outweigh them, according to Wang. “The significant risk reductions, particularly for cancers with high mortality rates, indicate a net positive effect,” she explained.

Another important wrinkle is that the study mostly covered the earliest eras of GLP-1 use. The first semaglutide-based drug, Ozempic, was only approved for type 2 diabetes in 2017, for instance, while tirzepatide was first approved in 2022. Given the higher effectiveness in weight loss seen with these newer medications, it’s certainly possible that they will have an even more profound effect in preventing cancer, Wang said.

On a sadder note, study author and cancer researcher Nathan Berger passed away just this past June at the age of 83. Berger was the founding director of the Case Comprehensive Cancer Center and more recently the director of the university’s Scientific Enrichment Opportunity (SEO) & Youth Engaged in Science (YES) programs that encourage high school students to get involved in scientific research. Berger was also one of Wang’s mentors.

“I would like to attribute this study to Dr. Nathan Berger, who conceptualized and supervised this study as the senior author.” Wang said.

As exciting as the idea that these increasingly popular drugs can also help prevent cancer might be, this research is only the beginning, according to Wang. We need studies that can directly examine this possibility as well as studies that can try to figure out exactly how these drugs fend off cancer.

“Moving forward, prospective randomized clinical trials are essential to confirm these associations and establish causality. Preclinical studies are needed to elucidate the underlying mechanisms,” she said. “Additionally, research to examine the newer and more potent weight loss GLP-1RAs (semaglutide and tirzepatide) would be particularly important.”

Pasteurization Kills Bird Flu Virus in Milk, New Studies Confirm

By Lauren J. Young

Pasteurization—the process of using high heat to eliminate harmful microbes in foods—effectively kills the H5N1 avian influenza virus that is currently circulating in U.S. cattle. In a new study scientists at the Food and Drug Administration and U.S. Department of Agriculture pasteurized raw milk inoculated with high concentrations of the pathogen and found that the treatment inactivated it in all samples. The finding validates that this century-and-a-half-old technique can protect humans from infectious microbes. Both agencies recommend that people should not consume raw (unpasteurized) milk or raw milk products.

Concerns over the safety of dairy products have risen with reports of bird flu outbreaks at dairy cow farms in multiple states. As of June 28, there have been three human cases of avian flu in the U.S. this year, according to the Centers for Disease Control and Prevention. All identified cases were in dairy farm workers who had direct contact with infected cattle. The FDA and USDA had previously sampled retail dairy products that had been commercially pasteurized, about 20 percent of which contained H5N1 viral particles. None of the 297 samples—which came from 132 processing sites in 38 states and were collected from April 18 to 22—contained live infectious virus, however. (The results of the sampling study were initially released in May, prior to its upcoming publication in the Journal of Virology.)

The most common commercial dairy treatment process is high-temperature short-time (HTST) pasteurization, also called “flash pasteurization,” which involves rapidly churning raw milk through large tanks at a temperature of at least 161 degrees Fahrenheit (72 degrees Celsius) for approximately three to 15 seconds before it’s cooled. This kills most harmful pathogens, including bacteria and viruses. The new pasteurization study, which the FDA released on its site on June 28, replicated real-world pasteurization conditions to see how effective the process is at killing the currently circulating H5N1 strain in milk. The paper has been submitted to the Journal of Food Protection but has not yet been published.

---

On supporting science journalism

If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.

---

The researchers collected raw milk samples from farms in states with known outbreaks. The average concentration of viable H5N1 in positive samples was about 3,000 viral particles per milliliter. The team then artificially contaminated homogenized, raw whole milk with even higher viral concentrations—about five million viral particles per milliliter—and treated the samples in a HTST continuous-flow pasteurization system similar to those used by the dairy industry. The system pumped the milk through two heating chambers: the first preheated it to 100 degrees F (38 degrees C), and the second brought it to 162.5 degrees F (72.5 degrees C) before holding it in a tube at 161 degrees F (72 degrees C). After about 15 seconds of that high heat, the liquid was transferred to a separate tank to cool.

Pasteurization completely inactivated the virus in all of the tested samples. This was true even when the team checked on the milk midway through the process, before it went into the holding tube. The researchers predict that pasteurization may be effective at killing as many as one trillion viral particles per milliliter.

“Validating the effectiveness of the pasteurization parameters critically demonstrates that commercial milk processing is capable of controlling the … virus and further provides broad assurance that pasteurized milk and dairy products made from pasteurized milk are safe,” said Nathan Anderson, a co-author of the study and director of the Division of Food Processing Science and Technology at the FDA’s Center for Food Safety and Applied Nutrition, in a recent press release.

At last week’s meeting of the CDC’s Advisory Committee on Immunization Practices, agency researchers presented an update on the state of avian influenza surveillance in the U.S. Data so far indicate that the risk of H5N1 infection in the general public is low. But the CDC recommends people who come into close, prolonged contact with infected animals should take safety precautions, such as wearing a face mask, a face shield or goggles and disposable coveralls. The agency says its scientists are actively monitoring the situation and collaborating with the FDA, the USDA and state public health departments. Currently H5N1 has been detected in cattle in 132 dairy herds across 12 states.

H5N1 isn’t a new threat; it’s been present in bird populations since 1996. The pathogen has since been detected in a growing number of mammals, including cats, dogs, goats and seals. The currently circulating version of the virus emerged in 2020 from migratory wild aquatic birds, first resulting in outbreaks in poultry in 2022 before being detected in cattle this year. Avian flu viruses have caused periodic outbreaks in humans, with a high mortality rate. The current strain does not yet appear to transmit easily from person to person, but the more it spreads, the greater the chances are that it will adapt to humans and cause a wider outbreak.

Liz González’s neighborhood in East San Jose can be loud. Some of her neighbors apparently want the whole block to hear their cars, others like to light fireworks for every occasion, and occasionally there are gunshots.

In February 2023, San Jose began piloting AI-powered gunshot detection technology from the company Flock Safety in several sections of the city, including Gonzalez’s neighborhood. During the first four months of the pilot, Flock’s gunshot detection system alerted police to 123 shooting incidents. But new data released by San Jose’s Digital Privacy Office shows that only 50 percent of those alerts were actually confirmed to be gunfire, while 34 percent of them were confirmed false positives, meaning the Flock Safety system incorrectly identified other sounds—such as fireworks, construction, or cars backfiring—as shooting incidents. After Flock recalibrated its sensors in July 2023, 81 percent of alerts were confirmed gunshots, 7 percent were false alarms, and 12 percent could not be determined one way or the other.

For two decades, cities around the country have used automated gunshot detection systems to quickly dispatch police to the scenes of suspected shootings. But reliable data about the accuracy of the systems and how frequently they raise false alarms has been difficult, if not impossible, for the public to find. San Jose, which has taken a leading role in defining responsible government use of AI systems, appears to be the only city that requires its police department to disclose accuracy data for its gunshot detection system. The report it released on May 31 marks the first time it has published that information.

The false-positive rate is of particular concern to communities of color, some of whom fear that gunshot detection systems are unnecessarily sending police into neighborhoods expecting gunfire. Nonwhite Americans are more often subjected to surveillance by the systems and are disproportionately killed in interactions with police. “For us, any interaction with police is a potentially dangerous one,” says Gonzalez, an organizer with Silicon Valley De-Bug, a community advocacy group based in San Jose.

San Jose did not attempt to quantify how many shooting incidents in the covered area the Flock System failed to detect, also known as the false-negative rate. However, the report says that “it is clear the system is not detecting all gunshots the department would expect.”

Flock Safety says its Raven gunshot detection system is 90 percent accurate. SoundThinking, which sells the ShotSpotter system, is the most popular gunshot detection technology on the market. It claims a 97 percent accuracy rate. But the data from San Jose and a handful of other communities that used the technologies suggest the systems—which use computer algorithms, and in SoundThinking’s case, human reviewers, to determine whether the sounds captured by their sensors are gunshots—may be less reliable than advertised.

Last year, journalists with CU-CitizensAccess obtained data from Champaign, Illinois, showing that only 8 percent of the 64 alerts generated by the city’s Raven system over a six-month period could be confirmed as gunfire. In 2021, the Chicago Office of Inspector General reported that over a 17-month period only 9 percent of the 41,830 alerts with dispositions that were generated by the city’s ShotSpotter system could be connected to evidence of a gun-related crime. SoundThinking has criticized the Chicago OIG report, saying it relied on “incomplete and irreconcilable data.”