@MSFTResearch: Securing data when trusted hardware is compromised; monitoring underwater recordings to protect endangered whales; Livi…

X AI KOLs Following News
data-security trusted-execution-environments distributed-consensus ai-audio-monitoring endangered-species museum-archives ai-ethics

Summary

Microsoft Research presents updates on Pirateship for securing data in compromised hardware, an AI system for monitoring endangered killer whales, and Living Library for interactive historical conversations, along with a call for AI builders to engage young users.

Securing data when trusted hardware is compromised; monitoring underwater recordings to protect endangered whales; Living Library enables conversations with historical figures; AI builders should listen to the young people who will ultimately use it. https://t.co/ZOMhhDnTmA https://t.co/jbsEJOuU0V
Original Article
View Cached Full Text

Cached at: 09/28/26, 05:34 PM

Securing data when trusted hardware is compromised; monitoring underwater recordings to protect endangered whales; Living Library enables conversations with historical figures; AI builders should listen to the young people who will ultimately use it. https://t.co/ZOMhhDnTmA https://t.co/jbsEJOuU0V

Monitoring endangered killer whales, bringing museum archives to life

Source: https://www.linkedin.com/pulse/monitoring-endangered-killer-whales-bringing-museum-djdve/ Microsoft Research

Microsoft Research

We advance science and technology to benefit humanity.

Published Sep 28, 2026

In this issue: Pirateship keeps data secure even when trusted hardware is compromised; AI monitoring of underwater recordings could help protect endangered killer whales; Living Library converts museum archives into interactive conversations with historical figures; Organizations building AI should listen to the people who will ultimately use it.

NEW RESEARCH

https://www.linkedin.com/redir/redirect?url=https%3A%2F%2Fwww%2Emicrosoft%2Ecom%2Fen-us%2Fresearch%2Fpublication%2Fpirateship-append-only-ledgers-for-mostly-trusted-execution-environments%2F&urlhash=xA1v&trk=article-ssr-frontend-pulse_publishing-image-blockMany industries rely on distributed ledgers to support collaboration while safeguarding sensitive data. Spreading trust across multiple administrative domains, which are unlikely to be compromised simultaneously, provides strong protection and high availability for critical data. These systems often combine cryptographic and algorithmic security with hardware-based security mechanisms, such as Trusted Execution Environments (TEEs). Unfortunately, TEEs can themselves be compromised, creating rare but potentially catastrophic failures.

This research presents Pirateship, a new distributed consensus protocol that cautiously trusts the guarantees of TEEs. Pirateship carefully embeds a Byzantine fault-tolerant (BFT) protocol inside of a crash-fault-tolerant (CFT) protocol without requiring additional messages. In representative deployments and applications, Pirateship achieves performance comparable to conventional TEE-enabled consensus protocols, while preserving data integrity even when TEE platforms are compromised.

NEW RESEARCH

https://www.linkedin.com/redir/redirect?url=https%3A%2F%2Fwww%2Emicrosoft%2Ecom%2Fen-us%2Fresearch%2Fpublication%2Fefficient-passive-acoustic-monitoring-of-killer-whales-using-a-two-stage-detection-and-ecotype-classification-cascade%2F&urlhash=P0qd&trk=article-ssr-frontend-pulse_publishing-image-blockA new AI system listens to continuous underwater recordings to detect killer whale vocalizations, and then identifies which of five eastern North Pacific ecotypes produced them. By providing faster, more reliable detections, the system could help inform efforts to protect endangered whales.

This system is designed to work in real time on low-power equipment, including edge devices deployed in the Puget Sound. It can identify whale vocalizations even amid heavy background noise and it filters out ambiguous calls that might otherwise result in incorrect classifications.

The conservation payoff is particularly important for the endangered Southern Resident killer whales. Knowing when those whales are present can help inform seasonal management measures such as vessel speed and approach restrictions.

NEW RESEARCH

https://www.linkedin.com/redir/redirect?url=https%3A%2F%2Fwww%2Emicrosoft%2Ecom%2Fen-us%2Fresearch%2Fpublication%2Fthe-living-library-transforming-archival-collections-into-conversational-knowledge-systems-lessons-from-the-theodore-roosevelt-presidential-library%2F&urlhash=UXJs&trk=article-ssr-frontend-pulse_publishing-image-blockThis research presents the Living Library, a system for converting fragmented digital archives into organized, interactive experiences for museum visitors. Developed and deployed at the Theodore Roosevelt Presidential Library, this system creates new ways for people to access, discover, and engage with history.

The Living Library includes four layers. The first three layers digitize archival materials, use AI to enrich and contextualize them, and organize the resulting collection so it can be searched and explored. The fourth layer turns the archival collection into an interactive, conversational experience, where visitors can ask questions of a life-size digital representation of Theodore Roosevelt. The authors hope their work helps museums and libraries activate their collections to provide a richer experience for the public.

NEW RESEARCH

https://www.linkedin.com/redir/redirect?url=https%3A%2F%2Fmicrosoft%2Egithub%2Eio%2Fnfw-reader%2Fposts%2Fhalf-finished-workplace&urlhash=y6_e&trk=article-ssr-frontend-pulse_publishing-image-blockThe next generation of workers is increasingly anxious about AI’s role in the workplace. Many young people are worried about inheriting workplaces that were designed without their input. In May 2026, nearly 40 NYU graduate students tested an early-stage AI prototype from Microsoft’s Office of Applied Research. Their task was not just to learn how to use it, but to help decide what it should become.

They found that the system could be swayed by a single confident assertion—even when the group disagreed. The experiment showed that AI that supports teamwork must understand who says something, how authoritative they are, and how decisions are made. This experience also gave these students a sense of agency. The researchers argue that organizations building or deploying AI should listen to the people who will ultimately use it.

Research Focus

Research Focus

150,864 followers

More articles by Microsoft Research

Explore content categories

Similar Articles