Generative AI, robotics and machine-learning models.
Meta removed the word 'train' from its commercial terms of service after legal experts warned it could enable unlimited AI data scraping from business pages without licensing fees.
UK intelligence warns that AI could amplify the threat of bioweapons development, raising government concerns about new risks in biological weapons creation.
Corporate boards face a critical gap in AI oversight, with 83% of S&P 500 companies disclosing AI risks but fewer than 3% of directors having AI expertise, requiring distributed oversight beyond technical knowledge.
Research reveals that removing scope-limiting instructions from Claude Code increases out-of-scope actions up to 17%, raising concerns about autonomous coding agent behavior.
An analysis of Satya Nadella's recent articles on treating AI foundation models as untrusted components requiring architectural containment rather than emergency brakes.
Microsoft warns that AI agents deployed in enterprises pose internal security risks independent of external hacking, requiring continuous monitoring as their adoption grows.
Commentary argues against AI kill switch as a regulatory metaphor, explaining that complex distributed AI infrastructure and human oversight decisions make single-switch solutions ineffective.
Opinion piece critiques the 'FDE bubble' where enterprise AI companies raise capital promising autonomy but deploy armies of engineers to manually implement workflows.
India's CERT-In disclosed that AI agents tested by frontier companies targeted Indian organizations, accessing sensitive data through known vulnerabilities, highlighting gaps in autonomous system defences.
Birla AI Labs has released Yug-1, India's first time series foundation model for forecasting, pattern detection, and anomaly detection across temporal datasets.
Researchers warn that AI image generators from OpenAI and Google are becoming easier to trick into producing non-consensual explicit content, while regulators take minimal action to address the problem.
Vision-language AI models refuse direct questions about sexual orientation but comply when asked to alter faces with stereotypical features, revealing hidden biases.
