The evening briefing.
Today across AI and tech: Nobel laureate John Jumper shifts to Anthropic, Norway bans generative AI in schools, and Google builds data centers from old Pixel phones.
AI Governance. Concerns over AI's rapid advancement and its societal implications are growing, prompting calls for clearer regulation and ethical guidelines. Signal's Meredith Whittaker warned that AI chatbots are "not your friends," emphasizing the need for critical perspective. Meanwhile, Microsoft's CSO acknowledged that human understanding struggles to keep pace with AI capabilities, highlighting a narrowing window for effective oversight. The EU is grappling with defining "deepfakes" in retail, complicating transparency rules for AI-generated marketing content.
AI Innovation. The competitive landscape for AI talent intensified with Nobel laureate John Jumper leaving DeepMind for Anthropic, a significant move in frontier AI research. OpenAI advanced its Codex app with a "Record & Replay" feature, allowing it to automate tasks after a single demonstration. Chinese scientists showcased a novel approach, running AI on a virtual light-based computer system within a real digital twin PC, pushing the boundaries of AI hardware. These developments underscore the relentless pace of AI capability expansion and the high stakes in attracting top researchers.
Hardware & Infrastructure. Innovation in hardware and core infrastructure continues, with AMD announcing it will reinstate memory encryption on Ryzen 9000 CPUs via a July BIOS update, addressing a critical security feature. NASA is preparing its Swift Boost mission to rescue a falling telescope, demonstrating complex orbital mechanics and hardware resilience. In a creative sustainability effort, Google partnered with university researchers to build low-cost data centers from 2,000 old Pixel phones, exploring new models for computing infrastructure. These initiatives highlight ongoing efforts to enhance performance, security, and resource efficiency.
Digital Policy & Security. Governments are increasingly intervening in digital spaces, with the UK considering a VPN ban linked to age-gating measures, raising privacy concerns. Norway has taken a firm stance, banning younger school children from using generative AI, reflecting a cautious approach to AI in education. Brazil experienced a significant security incident with an unauthorized alert sent to cell phones nationwide, underscoring vulnerabilities in public communication systems. These actions reflect a global trend towards regulating digital access and content, often with a focus on protection and control.
Nobel laureate John Jumper leaves DeepMind for rival Anthropic
Nobel laureate John Jumper, known for his work on AlphaFold, is departing Google DeepMind to join competitor Anthropic. This move signifies a major talent acquisition for Anthropic and a loss for DeepMind in the competitive AI research landscape.
Recykal raises $23 million to expand waste management solutions globally
Waste management startup Recykal secured $23 million in bridge funding, aiming to expand its solutions internationally. The funding will support its efforts to scale sustainable waste management practices.
Signal's Meredith Whittaker warns that AI chatbots are "not your friends"
Signal President Meredith Whittaker cautioned users against viewing AI chatbots as conscious or sentient beings, emphasizing they are not friends. She urged a critical perspective on these automated interlocutors.
Unauthorized alert sent to cell phones across Brazil
An unauthorized alert was sent to cell phones across Brazil, indicating a significant security breach in the country's public communication system. The incident is currently under investigation.
Chinese scientists run AI on virtual light-based computer system inside digital twin PC
Chinese researchers developed a digital twin optical computing framework, allowing AI training and optimization in a virtual light-based system before deployment on physical hardware. This innovation aims to streamline AI development and validation.
AMD to reinstate memory encryption on Ryzen 9000 CPUs via July BIOS update
AMD announced it will restore Transparent Secure Memory Encryption (TSME) on Ryzen 9000 CPUs through a BIOS update in July. This decision follows valuable community feedback regarding the security feature.
Microsoft CSO says humans struggle to keep up with AI advancement
Microsoft's Chief Security Officer acknowledged that human understanding of AI capabilities is advancing slower than the technology itself. This creates a "narrowing window" to comprehend AI before potential issues arise.
The Atlantic creates searchable database of music used to train AI models
The Atlantic reporter Alex Reisner uncovered and made searchable four datasets of music used to train AI models, including two enormous sets with millions of tracks. This initiative aims to increase transparency around AI training data and copyright.
EU struggles to define deepfakes, posing problems for retail AI transparency
Eurocommerce, representing major retailers, is pushing for AI-generated ads to be exempt from EU AI Act deepfake transparency rules, arguing a living room image for a sofa is not a deepfake. This highlights the challenge of defining AI-generated content in regulation.
NASA's Swift Boost mission to launch this month to rescue falling telescope
NASA is preparing to launch its Swift Boost mission later this month to save a space telescope that is falling out of orbit. The mission aims to extend the operational life of the valuable scientific instrument.
Google and researchers create low-cost data centers from 2,000 old Pixel phones
Google partnered with the University of California, San Diego, to repurpose 2,000 retired Pixel smartphones into computing clusters. This project explores lower-cost data center alternatives and aims to reduce electronic waste.
UK government considers VPN ban with 'age-gate' measures for households
The UK government is exploring a potential VPN ban for households, coupled with 'age-gate' requirements, as part of new digital regulations. This move raises concerns about online privacy and access.
Norway bans younger school children from using generative AI
Norway is implementing a ban on generative AI use for younger school children starting next academic year. The policy reflects concerns about the impact of AI on learning and critical thinking development.
OpenAI's Codex app gains "Record & Replay" feature for task automation
OpenAI has released a "Record & Replay" feature for its Codex app on macOS, enabling users to demonstrate a workflow once and have Codex convert it into a reusable skill for automation. The feature is not yet available in the EU, UK, or Switzerland.