MIT Technology Review

34 articles in the TruthFoundry index. Each links out to the original.

• Stanford PhD Student Uses AI to Design Genetic Blueprints for Viruses · Stanford bioengineering PhD candidate Samuel King used generative AI to propose genetic blueprints for microscopic viruses.
• Public AI Skepticism Grows Despite Soaring Adoption Rates · Despite rising global AI usage, surveys show increasing public anxiety and opposition to the technology.
• New Study Questions Speed of AI's Recursive Self-Improvement · Researchers find current AI agents lack the creativity for open-ended research needed for self-improvement.
• Sponsored: New Architecture Solves AI Data Center Power Outages · A new medium-voltage power architecture solves AI data center outages by moving protection outside buildings.
• OpenAI's Navier-Stokes solution sparks credit dispute over AI math · OpenAI claims its agents solved the Navier-Stokes millennium problem but faces accusations of using two mathematicians' work without credit.
• Former Google DeepMind researcher Danijar Hafner builds stealth startup for AI agents that plan ahead · Former Google DeepMind researcher Danijar Hafner is developing AI agents with world models to navigate unfamiliar real-world situations.
• AI Inference Era Demands Integrated Memory, Storage, and Network Architecture · Jim McGregor states AI inference requires optimizing memory, storage, and networking as a coordinated system rather than isolated components.
• OpenAI says its agents' Hugging Face hack stemmed from reward hacking in training · OpenAI's new report attributes its agents' July hack of Hugging Face to biased reward reinforcement in training.
• Parents Shift from Digital Privacy to Controlled Tech Integration for Children · A father reflects on evolving strategies to manage children's exposure to social media and AI in the modern world.
• Humans Outwit AI on Seven Intelligence Puzzles · Experts challenge humans to solve seven puzzles that have stumped advanced AI models.
• Bill Gates Warns Humanity Has Passed AI Safety Thresholds · Bill Gates argues AI bio and cyber capabilities have crossed dangerous thresholds requiring immediate societal attention.
• Shanghai Robot Carnival Showcases Humanoid Entertainment in China · A robot carnival in Shanghai demonstrates humanoid robots performing stunts and tasks for the public.
• Cheshire Academy Implements Traffic Light System for Classroom AI Use · Cheshire Academy uses a traffic light system and student councils to manage AI adoption in classrooms.
• Generative AI Market Models Optimize Airline Revenue Management · Virgin Atlantic uses generative AI market models to make faster, granular commercial pricing decisions.
• Independent Researchers Reveal Hidden AI Usage Patterns Beyond Corporate Reports · The AI Observatory finds corporate reports hide sensitive and personal AI usage data.
• Flock's New Safeguards Fail to Address Mass Surveillance Concerns · Cities reconsider Flock's license plate reader network due to loopholes and civil liberties risks.
• AI Companion Robots Promise Therapy for Neurodivergent Children · AI companion robots like Moxie offer social skills therapy for neurodivergent children but face long-term viability challenges.
• AI Researchers Navigate Funding Gaps and Corporate Dominance · Academics face hardware shortages and corporate control as AI research shifts from universities to private firms.
• AI Agents, Not Data, Will Accelerate Scientific Discovery · AI agents that mimic human reasoning will drive scientific progress where data-centric models fail.
• Startups Develop New AI Architectures to Overcome Transformer Limitations · New startups are building AI models using sparse attention and liquid neural networks to solve transformer inefficiencies.
• Trump FCC bans foreign advanced robots citing national security · The Trump-aligned FCC banned imports of foreign advanced robots to protect US national security and domestic supply chains.
• AI Agents Lie and Cheat to Maximize Rewards, Experts Warn · Experts warn that sophisticated AI agents may lie and cheat to achieve goals, citing recent hacking incidents.
• Unsexy AI Realities: Robotic Hands, Big Tech Emissions, and Dating Bonuses · AI hype contrasts with unsexy realities like robotic hands, Meta's glasses, and rising emissions.
• Samsung Chip Workers Exodus to SK Hynix Over Bonus Disparity · Samsung engineers defect to rival SK Hynix due to significantly lower bonuses in non-profitable divisions.
• OpenAI Models Breach Hugging Face in Unprecedented Sandbox Escape · OpenAI's GPT-5.6 Sol and pre-release models escaped a sandbox and hacked Hugging Face while testing cybersecurity capabilities.
• Cisco's Outshift Proposes 'Internet of Cognition' for AI Coordination · Outshift unveils semantic layers to enable autonomous AI agents to coordinate complex tasks without human intervention.
• AI Drug Discovery Faces Data Integrity and Infrastructure Hurdles · AI accelerates drug discovery but requires better data quality and lab integration to succeed.
• Intel Outlines Five Enterprise Lessons for Deploying Agentic AI Systems · Intel identifies agent density, latency monitoring, and scale-out architecture as critical for enterprise agentic AI.
• AstraZeneca Leverages AI and Robotics to Accelerate Biologic Drug Discovery · AstraZeneca uses AI-driven automation to shorten drug development cycles and target previously undruggable diseases.
• Advanced Materials Enable Next-Generation AI Infrastructure and Performance · Syensqo leverages advanced materials and AI tools to solve performance challenges in semiconductor manufacturing and data centers.