When generative AI first moved from research labs into real-world business applications, enterprises made a tacit bargain: “Capability now, control later.” Feed your proprietary data into third-party AI models, and you will get powerful results. But your data passes through systems you do not own, under governance you do not set. The protections you rely on are only as durable as the provider’s next policy update.
Now, with generative AI established in everyday business operations and sophisticated new agentic AI systems advancing every day, companies are reevaluating the terms of that deal.
“Data is really a new currency; it’s the IP for many companies,” says Kevin Dallas, CEO of EDB, echoing a recurrent anxiety from customers. “The big concern is, if you’re deploying an AI-infused application with a cloud-based large language model, are you losing your IP? Are you losing your competitive position?”
That question is now fueling a movement toward reclaiming both the data and AI systems that have rapidly become part of core business infrastructure. AI and data sovereignty, which refers to breaking dependence on centralized providers and establishing genuine control over models and data estates, it is an urgent priority for many companies, says Dallas, citing internal EDB data: “70% of global executives believe they need a sovereign data and AI platform to be successful.”
The idea of AI sovereignty is becoming a global policy conversation. NVIDIA CEO Jensen Huang recently spoke about the need for such a shift at the World Economic Forum’s annual meeting at Davos in January 2026: “I really believe that every country should get involved to build AI infrastructure, build your own AI, take advantage of your fundamental natural resource—which is your language and culture—develop your AI, continue to refine it, and have your national intelligence be part of your ecosystem.”
This report explores how enterprises are pursuing sovereignty over their models and data estates in an era of rapid AI adoption. Drawing on a survey conducted by EDB of more than 2,050 senior executives and a series of interviews with industry experts, the research confirms that the sovereignty movement on the enterprise level is already well underway.
This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff. It was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.
Get your daily dose of health and medicine every weekday with STAT’s free newsletter Morning Rounds. Sign up here.
Good morning. STAT’s Helen Branswell was recently interviewed for Columbia Journalism Review about the challenges of reporting on the hantavirus outbreak. If you need a dose of empathy for journalists — or if you’re a reporter yourself, looking for guidance from the best — check it out. And if you want to subscribe to STAT+ afterward, you can do that here.
The Trump admin’s wish list for its next FDA leader
The Trump administration is working to identify the next FDA commissioner after Marty Makary resigned on Tuesday. The dream candidate is someone who can rebuild trust with agency staff, focus on the agency’s food policy, and continue to drive drug-approval reforms, STAT’s Daniel Payne and Lizzy Lawrence report.
Anxiety disorders are associated with impaired inhibitory neurotransmission mediated by γ-aminobutyric acid type A (GABA-A) receptors. Although benzodiazepines remain effective anxiolytics, their clinical utility is limited by sedation, cognitive impairment, tolerance, and dependence, prompting the search for mechanistically distinct GABAergic modulators. Among cannabinoid-related molecules, the strongest evidence for direct GABA-A receptor modulation concerns the endocannabinoid 2-arachidonoylglycerol (2-AG), which potentiates recombinant human α1β2γ2 receptors through residues located in the M4 helix of the β2 subunit. Here, we review the structural architecture, biophysical properties, and pharmacological profile of the human GABA-A α1β2γ2 isoform as the relevant molecular framework for evaluating this mechanism, while discussing the broader relevance of cannabinoid-related ligands and selected phytocannabinoids without assuming mechanistic equivalence. We further assess the hypothesis that 2-AG reaches the β2-M4 site through a membrane-access route and identify five conceptual barriers that currently limit translation of this mechanism into anxiolytic drug development: supraphysiological effective concentrations, unresolved synaptic-versus-extrasynaptic actions, uncertain subtype selectivity, incomplete validation of lipid-environment effects, and lack of clinical evidence linking this mechanism to anxiolysis in humans. We conclude that direct modulation through β2-M4 defines a mechanistically intriguing allosteric pathway distinct from benzodiazepine action; however, its location on a shared β2 subunit and the micromolar concentrations required for modulation represent substantial obstacles to the rational design of anxioselective agents based on this mechanism.
Background: Problematic use of digital technology has increased across the world. Despite growing research, evidence on treatment effectiveness across digital behaviors remains fragmented. Objective: This study aimed to systematically evaluate and compare the effectiveness of therapeutic interventions targeted at problematic use of digital technology across various behavioral domains. Methods: A systematic review and meta-analysis was conducted in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines (PROSPERO: CRD420251052442). Electronic searches of PubMed, Scopus, and Embase (up to April 2025) were conducted. It identified 125 eligible studies, including 73 randomized controlled trials (RCTs), 32 non-RCTs, 14 pre-post studies, and 6 pilot studies. The interventions that were assessed in these studies included psychological therapies, digital or web-based programs, exercise-based interventions, pharmacological treatments, neuromodulation, parent-focused programs, virtual reality–based interventions, educational programs, and multicomponent approaches. Random-effects meta-analyses using standardized mean differences (SMDs) were performed. Results: For problematic internet use, psychological treatments showed a strong effect (effect size=−2.68; <.001). Digital interventions also showed significant benefit (effect size=−1.16; <.001). For smartphone addiction, psychological treatments (effect size=−1.49; <.001) and exercise-based programs (effect size=−3.07; =.001) showed significant improvement. For gaming disorder, psychological treatments showed improvement (effect size=−1.01; =.02), but results were mixed. There were limited studies to calculate pooled results for social media addiction, pornography use, gambling, screen time, and over-the-top content watching. No treatment studies were found for problematic over-the-top content watching. High heterogeneity and evidence of small-study effects were observed in several studies. Conclusions: Overall, structured psychological therapies showed the most consistent benefit. These findings support structured interventions that aim for control of use and reduce cues linked to high use. Evidence remains limited for several emerging digital behaviors. More high-quality studies are needed in clinical settings and for less-studied forms of digital addiction.
Get your daily dose of health and medicine every weekday with STAT’s free newsletter Morning Rounds. Sign up here.
Good morning. My co-workers and pals Isabella Cueto and Lev Facher have been talking about alcohol for years. As STAT’s reporters on chronic disease and addiction, respectively, it’s right at the intersection of their beats, yet rarely covered as a public health issue. I’m happy to share that all their talking turned to reporting, and now an incredible series. The first parts are up now. Scroll down or skip ahead to start reading.
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.
Three things in AI to watch, according to a Nobel-winning economist
A few months before he won the Nobel Prize in economics in 2024, Daron Acemoglu published a paper that earned him few fans in Silicon Valley. He argued that AI would give only a small boost to US productivity and would not eliminate the need for human work.
Two years later, Acemoglu’s measured take has not caught on. The technology has advanced quite a bit since his cautious predictions, but the data is still largely on his side.
This story is from The Algorithm, our weekly newsletter giving you the inside track on all things AI. Sign up to receive it in your inbox every Monday.
The case for fixing everything
Stewart Brand, the counterculture icon and tech industry legend, considers maintenance a “civilizational” act. His new book argues that taking responsibility for maintaining something, whether a motorcycle, a monument, or the planet, can be radical.
Brand argues that maintainers haven’t gotten the laurels they deserve—and he’s right. Yet his vision of maintenance often feels solitary: profound, but more about personal fulfillment than tending to a shared world or making it better.
Lee Vinsel is an associate professor of science, technology, and society at Virginia Tech, a cofounder of The Maintainers, and the host of Peoples & Things, a podcast about human life with technology.
This story is from the latest edition of our print magazine, which is all about nature. Subscribe now to read the full issue and receive future print copies once they land.
The must-reads
I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.
1 The first zero-day exploit built by AI has been discovered Google spotted and stopped the attempted “mass exploitation event.” (CNBC) + The hackers used AI to discover an unknown bug. (NYT $) + AI-powered hacking has exploded into an industrial-scale threat. (Guardian) + New tools are simplifying online crime. (MIT Technology Review)
2 OpenAI just launched its answer to Claude Mythos Daybreak patches vulnerabilities before attackers find them. (The Verge) + Sam Altman said it will “continuously secure software.” (Gizmodo) + It will rival Anthropic’s Claude Mythos, which arrived a month ago. (BBC) + OpenAI is allowing wider access to its cyber models than Anthropic. (CNBC)
3 Trump is heading to China to spread the gospel of American tech While taking cues from Beijing’s more stringent approach. (Guardian) + But investors want Trump and Xi to stay out of AI’s way. (Reuters $) + Elon Musk and Tim Cook are joining him on the trip this week. (BBC)
4 Ilya Sutskever has testified on Sam Altman’s “pattern of lying” OpenAI co-founder Sutskever took the stand in the Altman v. Musk trial. (BI) + He said he spent a year gathering proof of Altman’s dishonesty. (Reuters $) + But he also added to OpenAI’s defense. (Wired $) + While Satya Nadella called attempts to remove Altman “amateur city.” (FT $) + Here’s what happened last week in the trial. (MIT Technology Review)
5 A new hantavirus vaccine is in the works Moderna and Korea University are developing an mRNA vaccine. (Wired $) + Here’s what you need to know about the cruise ship outbreak. (MIT Technology Review)
6 Texas has sued Netflix over alleged data harvesting and “addictive” design AG Ken Paxton accuses Netflix of secretly collecting and selling user data. (Quartz) + And spying on children while deliberately fostering addiction. (Guardian)
7 A data center guzzled 30 million gallons of water—and no one noticed The curious case serves as a warning for other data center projects. (Ars Technica)
8 Europe is reportedly selling spyware to human rights abusers EU states allegedly sold the tech to countries violating rights. (Bloomberg $)
9 The US government’s AI vetting announcement has mysteriously vanished It had detailed a security test agreement with Google, xAI, and Microsoft. (Gizmodo)
10 Amazon staff are using AI for pointless tasks just to inflate usage scores In a bid to impress managers. (FT $) + An AI expert says we should stop using AI so much. (MIT Technology Review)
Quote of the day
“This is like the cheating husband complaining about the cheating wife.”
—Anupam Chander, a professor of law and technology at Georgetown Law School, tells the New York Times that Elon Musk’s hypocrisy over OpenAI becoming a for-profit company will undermine his courtroom battle with Sam Altman.
One More Thing
STUART BRADFORD
How sounds can turn us on to the wonders of the universe
For decades, astronomy has relied on visual information to make sense of the cosmos: images, charts, and graphs. Now, some researchers are trying something different: listening to the universe.
Using sonification, the process of turning information into sound, they’re helping blind and visually impaired researchers explore the cosmos—and even uncover patterns that might otherwise go unnoticed. The approach is spreading beyond astronomy into fields like climate science, navigation, and education.
Currently, formal models of addiction focus either on the complex individual decision-making processes involved in addiction or on the social dynamics of addiction. They do not integrate these two levels, which has been identified as a key shortcoming of current formal models of addiction. To address this, we propose a nonlinear dynamical modeling framework of addiction integrating both the individual level and social level of addictive behavior. The individual level of our modeling framework is a formalization of a dual-process theory, where one type of process increases the consumption of addictive goods, and another type of process limits consumption. For our formalization, we build on a well-studied model from ecology, originally used to model periodic outbreaks of the spruce budworm population. To this model, we add the process of incentive sensitization at the individual level and at the social level, we incorporate the critical processes of selection homophily and peer influence. We show that our integrated modeling framework can be used to explain key phenomena identified in addiction literature: a gradual transition to heavy use, sudden relapse and sudden quitting, relatively stable use states over time (i.e., abstinence moderate use, and heavy use), social contagion and sudden outbreaks, clustering of users, and social aid in recovery. In addition, we demonstrate how our modeling framework can be extended to include mutualistic, competitive, and more complex interactions between different addictive behaviors. Finally, we show how our framework can lead to new insights and predictions and suggest avenues for future research. (PsycInfo Database Record (c) 2026 APA, all rights reserved)
President Trump on Monday suggested the federal government could move to approve some forms of 7-OH, an opioid derived from the naturally occurring kratom plant.
“We’re looking very seriously at natural 7-OH and getting that approved,” Trump said.
It was not clear what Trump meant by “natural 7-OH.” Small amounts of the compound, shorthand for 7-hydroxymitragynine, occur naturally in kratom, which is increasingly used as a recreational drug and an unapproved pain treatment. While kratom is significantly less dangerous than potent synthetic opioids like fentanyl or prescription pain pills, it can still cause addiction and overdose.
<![CDATA[Psilocybin therapy shows fast, lasting relief for depression; clinicians discuss trial hurdles and emerging promise for PTSD and addiction in this podcast.]]>
Background: Substance use disorders (SUDs) are a major public health concern, contributing to significant individual and societal costs. Despite this, the uptake of evidence-based pharmacologic and behavioral interventions remains limited. The digital delivery of SUD treatment has emerged as a potentially scalable way to reduce access barriers and increase treatment use. Existing digital therapeutic interventions are often created without clinician involvement, evidence-based materials, interdisciplinary input, or content review. The implementation of a structured and methodologically rigorous development process is needed across digital health interventions to help ensure patient-facing materials are validated, understandable, and actionable for the end user. Objective: This early report seeks to describe and evaluate an iterative, interdisciplinary, platform-agnostic process for adapting and refining existing print materials for digital therapeutic modules in SUD treatment. The a priori goal was to evaluate if a structured, human-centered approach would generate digital modules that were rated as understandable and actionable based on a validated assessment for written materials. Methods: Fourteen therapeutic modules were adapted from existing Mayo Clinic–written, patient-facing education materials originally developed by a board-certified addiction psychiatrist and a doctoral-level education specialist for clinical use. A team of 4 purposively recruited licensed alcohol and drug counselors with lived experience with a SUD, all in recovery, and a doctoral-level therapeutic specialist met weekly for one hour over a 6-month period to iteratively adapt this existing content for smartphone delivery (2‐3 hours per module). The process flow included selecting source material, restructuring content for viewing on a phone screen, simplifying language, improving organization and flow to promote understanding, and including specific actions users could take based on the content. The counselors then independently evaluated the modules using the Patient Education Materials Assessment Tool for printable materials (PEMAT-P). PEMAT-P scores for understandability and actionability were calculated as percentages, and descriptive statistics were used to summarize scores in aggregate and across modules. A target of >70% was set for each PEMAT-P domain, consistent with accepted benchmarking standards. Results: Mean understandability and actionability for all modules were 87.2% (SD 4.8%; range 81.4%‐96.9%) and 75.1% (SD 12.3%; range 57.1%‐95.0%), respectively, exceeding the recommended threshold. While all modules were adequately understandable, 35.7% (5/14) scored below the actionability threshold. Conclusions: This early report highlights the value of a human-centered, iterative process for adapting therapeutic materials for digital delivery in SUD treatment. Although the modules performed well overall on PEMAT-P benchmarks, actionability was less consistent than understandability, and aggregate scores masked weaknesses in several individual modules. This indicates that a standardized process does not guarantee actionable material across all content types. Involving current patients in this process may improve the end product by incorporating a perspective that was previously missed.
