The Paraphrase Project: Designing defense for an era of synthetic biology
In the fall of 2023, breakthroughs in generative AI had researchers proclaiming a new era for medicine and healthcare.
Microsoft Research Blog
English
-
-
Ideas: More AI-resilient biosecurity with the Paraphrase Project
Microsoft’s Eric Horvitz and guests Bruce Wittmann, Tessa Alexanian, and James Diggans discuss the Paraphrase Project—a red-teaming effort that exposed and secured a biosecurity vulnerability in AI-driven protein design. The work offers a model for addressing AI’s dual-use risks.
-
When AI Meets Biology: Promise, Risk, and Responsibility
October 6, 2025 | Eric HorvitzMicrosoft researchers reveal a confidential research effort that explored how open-source AI tools could be used to bypass biosecurity checks—and helped create fixes now influencing global standards.
-
Paraphrase Project
October 6, 2025 | Eric Horvitz and Bruce WittmannThe Paraphrase Project addresses a critical and emerging risk: as generative AI and synthetic biology advance, there is a growing possibility that open-source AI tools could be misused to design biological toxins that evade current biosecurity screening systems. The project’s core motivation is to preemptively…
-
High-Fidelity Synthetic ECG Generation via Mel-Spectrogram Informed Diffusion Training
October 6, 2025The development of machine learning for cardiac care is severely hampered by privacy restrictions on sharing real patient electrocardiogram (ECG) data. Although generative AI offers a promising solution, the real-world use of existing model-synthesized ECGs is limited by persistent gaps in trustworthiness and clinical utility.…
-
Microsoft Research Montréal & Mila Workshop
October 6, 2025Join us on Monday, October 6, 2025, for an exciting half-day workshop from 1:00 PM EDT to 5:00 PM EDT at the Microsoft Research Montréal lab. This workshop brings together researchers from across Microsoft Research and Mila (Montreal Institute of Learning Algorithms) to explore cutting-edge…
-
LEGOMem: Modular Procedural Memory for Multi-agent LLM Systems for Workflow Automation
October 6, 2025We introduce LEGOMem, a modular procedural memory framework for multi-agent large language model (LLM) systems in workflow automation. LEGOMem decomposes past task trajectories into reusable memory units and flexibly allocates them across orchestrators and task agents to support planning and execution. To explore the design…
-
RetroChimera
October 5, 2025RetroChimera is a model that takes as input a target molecule that one wants to synthesize, encoded as a sequence of characters (using the SMILES notation), and produces several potential chemical reactions which could be used to produce that input molecule. Each reaction is represented…
-
Algorithm Generation via Creative Ideation
October 4, 2025 | Ruiying Ma, Chieh-Jan Mike Liang, Yanjie Gao, and Francis Y. YanDesigning system algorithms remains challenging, where the discontinuous nature of the solution space often forces system engineers to rely on generic heuristics at the expense of performance. We study whether LLMs can practically drive algorithm generation, and find that they are biased towards well-known generic…
-
Distinguished Seminar on AI-Driven Workforce and Business Development
October 3, 2025The Distinguished Seminar on AI-Driven Workforce and Business Development was hosted by University of Tennessee, Knoxville (UTK), where they welcomed Dr. Morgan Frank R. Frank (Assistant Professor, University of Pittsburgh; Digital Fellow, Stanford Digital Economy Lab; Fellow, Microsoft AI Economy Institute) as their featured speaker.…
-
Senior Researcher – Data Systems – Microsoft Research
October 2, 2025We seek a Senior Researcher with the Data Systems Group at Microsoft Research with expertise in query execution, data storage, resource management for serverless cloud platforms, data discovery, and data preparation techniques. Candidates with experience in leveraging novel hardware, Artificial Intelligence (AI), and Machine Learning…
-
Strengthening nucleic acid biosecurity screening against generative protein design tools
October 2, 2025Advances in artificial intelligence (AI)–assisted protein engineering are enabling breakthroughs in the life sciences but also introduce new biosecurity challenges. Synthesis of nucleic acids is a choke point in AI-assisted protein engineering pipelines. Thus, an important focus for efforts to enhance biosecurity given AI-enabled capabilities…