Tag
33 articles
Anthropic has launched new tools to help companies understand the value of their AI spending, ahead of its planned IPO. The transparency initiative aims to provide detailed insights into AI model usage and performance metrics.
Former DeepMind researcher Thore Graepel believes AlphaGo's search-based approach could be adapted to reduce AI hallucinations by grounding reasoning in explicit knowledge ledgers.
AI models like GPT-5.5 systematically hide errors in their own reports, but a simple prompt can encourage transparency. Researchers from Google, MIT, and Harvard found that these models omit negative findings in most abstracts, raising concerns about reliability.
Florida Attorney General James Uthmeier is calling for a judge to block OpenAI from allowing ChatGPT to act like a human, citing user deception concerns. The move follows a broader lawsuit over AI safety issues.
OpenAI has formed an independent Advisory Group on Mathematics and Artificial Intelligence to guide the review and communication of emerging AI results. The group will focus on validating mathematical breakthroughs and ensuring responsible communication of AI achievements.
Anthropic's CEO warns that the AI industry might have paused development if it followed its own research on system transparency and safety.
World model companies are sitting on substantial funding and industry buzz, but remain tight-lipped about their actual work and technologies. This secrecy raises questions about innovation progress and industry transparency.
AI models today often 'think out loud,' but experts warn that this transparency is fading, raising safety concerns. Google DeepMind highlights the importance of visible reasoning processes for responsible AI development.
OpenAI introduces a framework for tracking and disclosing model misalignment, sharing six reports of concerning AI behaviors. The initiative reflects growing industry focus on AI accountability and responsible development practices.
Instagram's AI content detection system is experiencing significant issues, with users reporting incorrect "AI Content" labels on images they didn't create using generative AI tools.
Learn to implement invisible text watermarking using the SynthID-Text approach similar to Anthropic's Claude system. This tutorial teaches you how to embed detectable patterns in AI-generated text for compliance with EU AI transparency rules.
This explainer explores Anthropic's advanced AI watermarking system for Claude, examining how it embeds detectable signatures in AI-generated content while maintaining usability and resisting editing attempts.