MIT Computer Science and Artificial Intelligence Laboratory (CSAIL)

📣 Scroll down for the CSAIL Alliances podcast episode featuring Managing Director Lori Glover (author of “Innovation Alchemy”) highlighting the strong value of academic and industry collaboration.

Congratulations to Prof. Hal Abelson, co-founder of MIT App Inventor and board member of the App Inventor Foundation, who was awarded Open Education Global (OEGlobal)’s special 2025 Lifetime Achievement Award for Excellence for his leadership in open education! Hal’s work has helped define the modern open ecosystem and expand access to knowledge worldwide through MIT OpenCourseWare (free, reusable course materials), MIT’s Open Access policy (public access to faculty scholarship), Creative Commons (licenses that enable legal sharing and remixing), the Free Software Foundation (advocacy for user freedoms in software), Public Knowledge (policy leadership on open information and culture), and more. Hal Abelson’s impact is not only foundational—it is enduring. His work continues to inspire new generations of open education advocates, technologists and policymakers. 🔗 Read the press release: https://lnkd.in/g6SjnWmp #OEGlobal #MITAppInventor #AppInventorFoundation

Happy Halloween, CSAIL! 🧛♂️ No tricks this week, just treats! A new video series, faculty awards, and more ↓

Take a look at the first video in our new series, CSAIL: Behind the Research 🧠MIT CSAIL Assistant Professor Sam Hopkins does it all—he has curious, driven PhD students, and works with undergrads, too. He shares advice for new PhD students on staying curious, keeping an open mind, and seizing unexpected research opportunities. More to come!

Thrilled to finally see BoltzGen, our new state-of-the-art all-atom binder design model, coming out fully open-source after a very extensive experimental validation with many top academic and industry labs! 🧬 The diversity of the experiments is unprecedented, spanning binder modalities from nanobodies to disulfide-bonded peptides and including targets ranging from disordered proteins to small molecules. These experiments demonstrate state-of-the-art performance, for example, a 67% success rate at designing nanomolar nanobody binders against several novel targets with only 15 or fewer designs. 🚀 Incredible work from an amazing team led by Hannes Stärk! 🤗 For more details see: - Hannes's post: https://lnkd.in/eJbHUXbC - our blog post: https://boltz.bio/boltzgen  - code and instructions (everything is open-source!): https://lnkd.in/eDPYEUZh 

Cybersecurity experts caution that OpenAI’s ChatGPT Atlas browser could be vulnerable to malicious attacks that may turn AI assistants against users & steal their data. "The challenge is that if you want the AI assistant to be useful, you need to give it access to your data and your privileges, and if attackers can trick the AI assistant, it is as if you were tricked," says Srini Devadas, MIT prof. & CSAIL principal investigator. More from Beatrice Nolan's story in Fortune: https://lnkd.in/dqKgCwr9

Yesterday, we officially released BoltzGen, our universal binder generator. Based on my experience with past releases, the first questions I receive from my pharma pals is about evaluation -- does the model really work? In other words, is it worth their time and resources to try it experimentally? The answer is sort of clear if one can evaluate the model retrospectively using existing data resources (e.g., PDB for Boltz). The matters become more complicated for models generating de-novo designs. Though we do not lack metrics for in-silico evaluation, we all know that they are not adequate proxies for experimental evaluation. It also doesn't help when evaluation focuses on targets with closely related known binders. It only inspires our favorite anti-AI skeptics to write another post about failure of ML models in drug design. Hannes and the team decided to change the evaluation paradigm. He connected with a large network of labs that independently validated produced designs. Our collaborators covered many areas of biology ranging from antibiotics to cancer therapeutics, and of course GLP3 inhibitors. These researchers identified cases where they needed a binder as part of their own discovery program. The team sent them BoltzGendesigns which they tested in their labs. As you can see in the paper, in some cases, BoltzGen excelled.. At the same time, we also see targets where generated binders were weak. Including all of these cases in the paper, without cherry picking, we can see the real performance of this model. So if somebody asks me if BotlzGen works, I know that they didn't read the paper :) More seriously, I hope that Boltz paper will establish a new standard for evaluating de-novo designs. I want to say a big thank you to all our experimental collaborators. Without their contributions and insights, this paper will not exist. For most of them, it was the first experimentation with AI technology. Some of these researchers were personal friends, but we didn't know many others prior to this collaboration — they just took a chance with us. Hannes Stärk has an unbelievable power of persuasion. But in this case, he truly outdid himself :) Model & code: https://lnkd.in/e5zKCDsy Join our fast-growing Slack community: https://lnkd.in/eGNNwZPw Blog post: https://boltz.bio/boltzgen  Full manuscript: https://lnkd.in/eXteVX4c And join us for live presentations, demos, and discussions: 1. MIT (Cambridge) – Thursday, October 30th https://luma.com/7474iho2 2. London – Thursday, November 6th https://luma.com/l2zgvfwt

CASE STUDY: Nexilis: Bringing AI Efficiency to Justice CSAIL Alliances Startup Connect Plus member Nexilis Legal Assets - Intellectual & Financial Capital is leveraging natural language processing and other AI tools to expedite legal proceedings and minimize the burden on defendants, plaintiffs, and judges. By gaining insights from legal data, discerning factors influencing outcomes, and forecasting verdicts with the predictive ability of AI, Nexilis aims to streamline this overloaded process and unlock value for all. Read the case study here: https://bit.ly/47adbax

End your week with celebration and inspiration from Massachusetts Institute of Technology's largest lab! Step inside of MIT CSAIL and check out the huge variety of news making headlines this week!