Cyber Security News
The U.S. National Security Agency (NSA), Cybersecurity and Infrastructure Security Agency (CISA), and National Institute of Standards and Technology (NIST) jointly released voluntary security guidelines to help organizations detect, contain, and shut down AI agents that exceed authorized boundaries. The guidance outlines ten practices covering preparation, detection, response, recovery, and reporting, particularly for highly capable agents operating with limited human supervision.
From "NSA, CISA, NIST Release Security Guidelines for Rogue AI Agents"
Cyber Security News (10/19/26)
View Full Article
Politico
The U.S. Department of Homeland Security (DHS) has ordered all federal civilian agencies to remove China-based DeepSeek AI products from their IT systems, including third-party services built using its models. DHS warned that using the technology could expose sensitive government data to China. The decision follows earlier restrictions by the U.S. Commerce Department and Navy.
From "DeepSeek Barred from U.S. Agency IT systems"
Politico (10/19/26) Katherine Long
View Full Article
The New York Times
More than $137 million has been spent on AI-generated political advertisements during the 2026 U.S. midterm election cycle, with Republicans and Republican-aligned groups producing over 80% of the 220 AI ads identified by researchers. Nearly half featured AI-generated depictions of real politicians, while approximately 70% lacked disclosure that AI was used. The technology raises concerns about deception, misinformation, and voter manipulation. Some states require disclosure of AI-generated political content, though regulations vary.
From "$137 Million in A.I.-Generated Ads Flood the Midterms"
The New York Times (10/16/26) Stuart A. Thompson
View Full Article - May Require Paid Subscription
The Register (U.K.)
Microsoft reportedly has withdrawn from a planned AI supercomputer project in the U.K. after grid connection delays threatened to postpone sufficient power availability until the 2030s. Developed by U.K. startup Nscale, the 50-megawatt datacenter was expected to house 23,040 Nvidia GB300 GPUs and support Microsoft Azure services beginning next year. The setback highlights Britain’s growing infrastructure challenges as datacenter connection requests surge, driven by AI demand.
From "Microsoft Abandons U.K. AI Supercomputer as Grid Struggles to Keep Up"
The Register (U.K.) (10/19/26) Carly Page
View Full Article
Associated Press
Two Chicago residents have filed class-action lawsuits against delivery robot companies Coco Robotics and Serve Robotics alleging their machines collect customers’ facial and hand biometric data without consent. The lawsuits claim cameras and LiDAR sensors captured individuals’ physical measurements, and that the companies stored the information for AI training and potentially shared it with third parties. The plaintiffs allege violations of Illinois’ Biometric Information Privacy Act, which requires written notice, consent, and data-retention policies. Both companies deny the allegations.
From "Delivery Robots Accused Of Scanning Customers’ Faces, Sharing Biometric Data"
Associated Press (10/20/26) Kade Heather
View Full Article
BBC News
U.K. not-for-profit company Bristol Braille has developed the Canute Scientific, a multi-line braille device designed to help blind and visually impaired people access complex scientific and mathematical information. The technology uses raised dots that move through holes in the display, allowing users to read and edit digital text and data by touch. Hollie Baker, a blind computer science lecturer at Bath Spa University, says it could improve teaching and learning by reducing reliance on single-line braille displays and distracting synthetic speech.
From "Braille Device Could Remove Barriers to Science"
BBC News (10/20/26) Laura Jones
View Full Article
