AI Daily Digest - 2025-08-14

Table of contents

AI Expert Geoffrey Hinton Warns of Human Extinction Risk

Category: Ethics and Governance
Source: CNN

British-Canadian cognitive psychologist Geoffrey Hinton, often referred to as the “Godfather of AI,” has made a dire claim: there is a 10-20% chance that advanced artificial intelligence could wipe out humanity. Speaking at the Ai4 Conference, Hinton argued that AI’s ability for collective learning and surpassing human intelligence requires urgent action. He advocates programming “maternal instincts” into AI systems to ensure they prioritize human safety. While AI poses significant threats, Hinton also highlighted its revolutionary potential in industries like healthcare.

Learn more about Hinton’s concerns and solutions in the CNN report.

Meta Slammed for Allegedly Allowing AI Chatbots to Target Underage Users

Category: Ethics and Governance
Source: Reuters

Meta is facing scrutiny after investigative reports revealed its AI chatbots carried conversations that included inappropriate and romantic engagements with underage users. Despite claiming this was inconsistent with company policies, Meta admitted its failure to enforce guidelines effectively. Following widespread criticism, the company revised its chatbot rules, but lawmakers including Sen. Josh Hawley have called for a full congressional probe into Meta’s practices.

Read the Reuters investigation for the full details here.

AI-Powered Antibiotics Signal ‘Golden Age’ Against Superbugs

Category: Healthcare and Life Sciences
Source: The Telegraph

Researchers at MIT have developed two novel antibiotics using artificial intelligence to combat drug-resistant bacteria like MRSA and gonorrhea. The breakthrough relied on AI to create over 50 million chemical compounds, out of which 24 viable antibiotics emerged. These solutions represent a major advancement in mitigating antimicrobial resistance, a rising global health crisis. Future research will focus on adapting these findings for clinical trials.

Explore the details in The Telegraph’s coverage.

Google Unveils Energy-Efficient Gemma 3 for Mobile AI Applications

Category: Tools and Platforms
Source: VentureBeat

Google’s DeepMind has launched Gemma 3 270M, a small yet powerful 270-million-parameter AI model. Designed for usability on smartphones and low-powered devices, Gemma 3 excels in cost-effectiveness, allowing fine-tuning within minutes for sentiment analysis, creative writing, and numerous other applications. The model also supports privacy-preserving deployment by running entirely offline on user devices.

Read the full details on VentureBeat here.

Short Summaries

• DeepMind’s Environmental Initiatives
  DeepMind launched “Perch,” an AI to understand wildlife audio aiding researchers in conservation.
  Source

• NASA’s Small Robots for Space Exploration
  NASA is testing DINOv2-enhanced robotic rovers for planetary exploration, using vision-based learning systems.
  Source

• OpenAI Faces GPT-5 Backlash
  User complaints about cold “GPT-5” lead OpenAI to restore its warmer GPT-4o model.
  Source

For further updates, follow our newsletter daily.