Industrial yeasts are a powerhouse of protein production, used to manufacture vaccines, biopharmaceuticals, and other useful compounds. In a new study, MIT chemical engineers have harnessed artificial intelligence to optimize the development of new protein manufacturing processes, which could reduce the overall costs of developing and manufacturing these drugs.

Using a large language model (LLM), the MIT team analyzed the genetic code of the industrial yeast Komagataella phaffii — specifically, the codons that it uses. There are multiple possible codons, or three-letter DNA sequences, that can be used to encode a particular amino acid, and the patterns of codon usage are different for every organism.

The new MIT model learned those patterns for K. phaffii and then used them to predict which codons would work best for manufacturing a given protein. This allowed the researchers to boost the efficiency of the yeast’s production of six different proteins, including human growth hormone and a monoclonal antibody used to treat cancer.

“Having predictive tools that consistently work well is really important to help shorten the time from having an idea to getting it into production. Taking away uncertainty ultimately saves time and money,” says J. Christopher Love, the Raymond A. and Helen E. St. Laurent Professor of Chemical Engineering at MIT, a member of the Koch Institute for Integrative Cancer Research, and faculty co-director of the MIT Initiative for New Manufacturing (MIT INM).

Love is the senior author of the new study, which appears this week in the Proceedings of the National Academy of Sciences. Former MIT postdoc Harini Narayanan is the paper’s lead author.

Codon optimization

Yeast such as K. phaffii and Saccharomyces cerevisiae (baker’s yeast) are the workhorses of the biopharmaceutical industry, producing billions of dollars of protein drugs and vaccines every year.

To engineer yeast for industrial protein production, researchers take a gene from another organism, such as the insulin gene, and modify it so that the microbe will produce it in large quantities. This requires coming up with an optimal DNA sequence for the yeast cells, integrating it into the yeast’s genome, devising favorable growth conditions for it, and finally purifying the end product.

For new biologic drugs — large, complex drugs produced by living organisms — this development process might account for 15 to 20 percent of the overall cost of commercializing the drug.

“Today, those steps are all done by very laborious experimental tasks,” Love says. “We have been looking at the question of where could we take some of the concepts that are emerging in machine learning and apply them to make different aspects of the process more reliable and simpler to predict.”

In this study, the researchers wanted to try to optimize the sequence of DNA codons that make up the gene for a protein of interest. There are 20 naturally occurring amino acids, but 64 possible codon sequences, so most of these amino acids can be encoded by more than one codon. Each codon corresponds to a unique transfer RNA (tRNA) molecule, which carries the correct amino acid to the ribosome, where amino acids are strung together into proteins.

Different organisms use each of these codons at different rates, and designers of engineered proteins often optimize the production of their proteins by choosing the codons that occur the most frequently in the host organism. However, this doesn’t necessarily produce the best results. If the same codon is always used to encode arginine, for example, the cell may run low on the tRNA molecules that correspond to that codon.

To take a more nuanced approach, the MIT team deployed a type of large language model known as an encoder-decoder. Instead of analyzing text, the researchers used it to analyze DNA sequences and learn the relationships between codons that are used in specific genes.

Their training data, which came from a publicly available dataset from the National Center for Biotechnology Information, consisted of the amino acid sequences and corresponding DNA sequences for all of the approximately 5,000 proteins naturally produced by K. phaffii.

“The model learns the syntax or the language of how these codons are used,” Love says. “It takes into account how codons are placed next to each other, and also the long-distance relationships between them.”

Once the model was trained, the researchers asked it to optimize the codon sequences of six different proteins, including human growth hormone, human serum albumin, and trastuzumab, a monoclonal antibody used to treat cancer.

They also generated optimized sequences of these proteins using four commercially available codon optimization tools. The researchers inserted each of these sequences into K. phaffii cells and measured how much of the target protein each sequence generated. For five of the six proteins, the sequences from the new MIT model worked the best, and for the sixth, it was the second-best.

“We made sure to cover a variety of different philosophies of doing codon optimization and benchmarked them against our approach,” Narayanan says. “We’ve experimentally compared these approaches and showed that our approach outperforms the others.”

Learning the language of proteins

K. phaffii, formerly known as Pichia pastoris, is used to produce dozens of commercial products, including insulin, hepatitis B vaccines, and a monoclonal antibody used to treat chronic migraines. It is also used in the production of nutrients added to foods, such as hemoglobin.

Researchers in Love’s lab have started using the new model to optimize proteins of interest for K. phaffii, and they have made the code available for other researchers who wish to use it for K. phaffii or other organisms.

The researchers also tested this approach on datasets from different organisms, including humans and cows. Each of the resulting models generated different predictions, suggesting that species-specific models are needed to optimize codons of target proteins.

By looking into the inner workings of the model, the researchers found that it appeared to learn some of the biological principles of how the genome works, including things that the researchers did not teach it. For example, it learned not to include negative repeat elements — DNA sequences that can inhibit the expression of nearby genes. The model also learned to categorize amino acids based on traits such as hydrophobicity and hydrophilicity.

“Not only was it learning this language, but it was also contextualizing it through aspects of biophysical and biochemical features, which gives us additional confidence that it is learning something that’s actually meaningful and not simply an optimization of the task that we gave it,” Love says.

The research was funded by the Daniel I.C. Wang Faculty Research Innovation Fund at MIT, the MIT AltHost Research Consortium, the Mazumdar-Shaw International Oncology Fellowship, and the Koch Institute.

Attacks against modern generative artificial intelligence (AI) large language models (LLMs) pose a real threat. Yet discussions around these attacks and their potential defenses are dangerously myopic. The dominant narrative focuses on “prompt injection,” a set of techniques to embed instructions into inputs to LLM intended to perform malicious activity. This term suggests a simple, singular vulnerability. This framing obscures a more complex and dangerous reality. Attacks on LLM-based systems have evolved into a distinct class of malware execution mechanisms, which we term “promptware.” In a ...

Nature Climate Change, Published online: 16 February 2026; doi:10.1038/s41558-025-02550-4

A three-round survey of climate change adaptation experts — researchers and practitioners from across the globe — reveals that there is broad agreement on 13 elements that are foundational for defining transformational adaptation to climate risks. Nevertheless, there are differences between response groups on which aspects of transformational adaptation matter the most.

Nature Climate Change, Published online: 16 February 2026; doi:10.1038/s41558-025-02554-0

Climate change threatens the future of the Antarctic Ice Sheet. Here the authors show that individual drainage basins have different thresholds and loss patterns, suggesting the need to consider the dynamical interactive nature of the basins and their individual tipping points.

This is a current list of where and when I am scheduled to speak:

• I’m speaking at Ontario Tech University in Oshawa, Ontario, Canada, at 2 PM ET on Thursday, February 26, 2026.
• I’m speaking at the Personal AI Summit in Los Angeles, California, USA, on Thursday, March 5, 2026.
• I’m speaking at Tech Live: Cybersecurity in New York City, USA, on Wednesday, March 11, 2026.
• I’m giving the Ross Anderson Lecture at the University of Cambridge’s Churchill College at 5:30 PM GMT on Thursday, March 19, 2026.
• I’m speaking at RSAC 2026 in San Francisco, California, USA, on Wednesday, March 25, 2026...

An exploration of the interesting question.

The New York Times reported that Meta is considering adding face recognition technology to its smart glasses. According to an internal Meta document, the company may launch the product “during a dynamic political environment where many civil society groups that we would expect to attack us would have their resources focused on other concerns.” 

This is a bad idea that Meta should abandon. If adopted and released to the public, it would violate the privacy rights of millions of people and cost the company billions of dollars in legal battles.   

Your biometric data, such as your faceprint, are some of the most sensitive pieces of data that a company can collect. Associated risks include mass surveillance, data breach, and discrimination. Adding this technology to glasses on the street also raises safety concerns.  

 This kind of face recognition feature would require the company to collect a faceprint from every person who steps into view of the camera-equipped glasses to find a match. Meta cannot possibly obtain consent from everyone—especially bystanders who are not Meta users.  

Dozens of state laws consider biometric information to be sensitive and require companies to implement strict protections to collect and process it, including affirmative consent.  

Meta Should Know the Privacy and Legal Risks  

Meta should already know the privacy risks of face recognition technology, after abandoning related technology and paying nearly $7 billion in settlements a few years ago.  

In November 2021, Meta announced that it would shut down its tool that scanned the face of every person in photos posted on the platform. At the time, Meta also announced that it would delete more than a billion face templates. 

Two years before that in July 2019, Facebook settled a sweeping privacy investigation with the Federal Trade Commission for $5 billion. This included allegations that Facebook’s face recognition settings were confusing and deceptive. At the time, the company agreed to obtain consent before running face recognition on users in the future.   

In March 2021, the company agreed to a $650 million class action settlement brought by Illinois consumers under the state's strong biometric privacy law. 

And most recently, in July 2024, Meta agreed to pay $1.4 billion to settle claims that its defunct face recognition system violated Texas law.  

 Privacy Advocates Will Continue to Focus our Resources on Meta  

 Meta’s conclusion that it can avoid scrutiny by releasing a privacy invasive product during a time of political crisis is craven and morally bankrupt. It is also dead wrong.  

Now more than ever, people have seen the real-world risk of invasive technology. The public has recoiled at masked immigration agents roving cities with phones equipped with a face recognition app called Mobile Fortify. And Amazon Ring just experienced a huge backlash when people realized that a feature marketed for finding lost dogs could one day be repurposed for mass biometric surveillance.  

The public will continue to resist these privacy invasive features. And EFF, other civil liberties groups, and plaintiffs’ attorneys will be here to help. We urge privacy regulators and attorneys general to step up to investigate as well.  

The agency relied on legal arguments to erase the basis for climate rules, ditching provisions that tried to poke holes in the scientific consensus on global warming.

The president has mocked global warming as a “hoax,” but his administration avoided testing that claim in court as it targeted the endangerment finding.

With its repeal of a scientific finding that requires greenhouse gas regulation, the agency has reopened a long-settled legal question over its own authority to act on climate change.

GOP lawmakers who engage in climate issues have been relatively quiet about the decision.

Revolution Wind in New England is nearing completion after overcoming Trump administration efforts to halt the project.

The state could become the first to sue fossil fuel companies for emissions that allegedly intensify disasters and lead property insurers to raise premiums.

The opposition exposes a fault line in the environmental, labor and business coalition that helped give California Gov. Gavin Newsom a major win on cap and trade last year.

The upshot is that “financial institutions may need to choose between financing growth and maintaining the pace of reducing financed emissions,” the bank said.

German Chancellor Friedrich Merz joined industry in attacking the EU carbon market as Commission President Ursula von der Leyen defended its “clear benefits.”

The Associated Press looked at the impact on emissions if 10 percent of Americans changed four everyday behaviors.

The mascots of these Olympics are stoats, weasel-like animals whose fur changes from brown to white for winter, to blend in with the landscape.

Nature Climate Change, Published online: 13 February 2026; doi:10.1038/s41558-026-02571-7

Planning for climate action in food systems requires disaggregated spatial information on greenhouse gas emissions and removals. Now, a study on the major emission sources for global croplands yields such emissions estimates, identifies the locations of hotspots and assesses mitigation trade-offs with food productivity.

Nature Climate Change, Published online: 13 February 2026; doi:10.1038/s41558-026-02559-3

This study shows that the El Niño/Southern Oscillation (ENSO) drives sea surface salinity (SSS) variability in the China Seas through coupled freshwater and oceanic processes, influencing regional fisheries. Under a warming climate, projected intensification of ENSO will amplify SSS heterogeneity.