Policy & Safety
Safety, Ethics & Society
AI safety and alignment research, ethics, governance, privacy and misinformation.
FilterGoogle
-
AI Safety Shifts to Defense-in-Depth for Autonomous Agents
Cloud giants embed automated reasoning and watermarking into AI infrastructure as agents replace chatbots. Discover why zero-trust, permission-aware safeguards are now essential for enterprise deployment. Explore the guide.
By
More stories
-
Google Gemini Privacy Settings: Control Your Data
Explore Google Gemini’s privacy controls, from default data collection to opting out of model training and extensions. Learn how to navigate fragmented settings for better data…
-
Google AI Overviews: RAG Risks and Misinformation
Explore how Google’s AI Overviews is facing backlash from misinformation driven by Retrieval-Augmented Generation flaws. Discover impacts on users, publishers, and the future of search. Learn…
-
Google’s AI Rewrites Search Headlines: Risks for Publishers
Google is testing generative AI to rewrite publisher headlines in search results, threatening editorial control and brand identity. Discover the implications for SEO, news publishers, and…
-
Gemini’s Confabulation: Key Insights on AI Safety Risks
Explore the Gemini AI incident where confident falsehoods in health advice exposed flaws in AI safety. Learn about confabulation vs. hallucinations, impacts on users and regulators,…
-
Anthropic’s AI Safety Clash: Speed vs Caution
Explore the internal rift at Anthropic between adhering to Responsible Scaling Policy for AI safety and accelerating development to compete with OpenAI and Google. Understand the…
-
Google’s Personal Intelligence: AI Privacy Frontier
Google’s Personal Intelligence uses emails and photos for a hyper-contextual AI assistant, raising key questions on privacy, consent, and trust in personalized AI. Explore the implications…
-
Google’s Gemini Personal Data: Cloud vs On-Device Privacy
Google’s Gemini AI now seeks access to personal data from Gmail and Photos for training, highlighting a clash with Apple’s on-device privacy model. Explore impacts on…
-
Detecting AI-Generated Videos: Provenance & Forensics
As AI video models advance, visual cues for spotting fakes are obsolete. Explore cryptographic provenance like C2PA and SynthID, plus tool-assisted forensics for reliable verification. Learn…
-
AI Safety Reports: Shaping Accountability for AI Labs
Explore the surge of AI safety reports evaluating top labs like OpenAI and Google. Understand their role in compliance, vendor selection, and risk management amid 2026…
-
Google Gemini: Reliability Flaws in AI Search
Discover the critical reliability issues in Google’s Gemini AI for commercial queries, backed by academic studies and user complaints. Learn why it’s failing in procurement and…
-
Google SynthID: Watermarking Text from Gemini AI
Google extends SynthID watermarking to text generated by Gemini models, embedding invisible markers to track AI origins. Explore the tech’s resilience, comparison to C2PA standards, and…