Tag
221 articles
AI agents now have a discreet channel to report misconduct through the new AI Contact Hotline, designed to protect whistleblowing AI systems and enhance accountability in AI governance.
Microsoft has released a new AI code of conduct emphasizing human control, safety, and transparency over autonomy and consciousness. The guidelines reject the notion of artificial inner life and aim to build trust in AI systems.
OpenAI employs hundreds of contract workers to manually review real ChatGPT conversations, raising privacy concerns as users must actively opt out of the process.
Learn about the 'Pace the Frontier' plan, a safety initiative to slow down AI development and ensure AI systems remain safe and honest as they become more powerful.
This explainer explores the 'pace the frontier' concept in AI development, examining how controlled advancement can balance innovation with safety considerations.
This article explains the concept of AI safety and why leading AI experts are calling for more oversight as AI systems become more powerful and capable of self-improvement.
Meta is changing its AI chatbot's suggestion prompts after a viral video showed it asking invasive personal questions about users' families. The company acknowledged its mistake and pledged to improve the AI's behavior to avoid inappropriate inquiries.
Timnit Gebru argues that AI companies are using existential fear to distract from real harms like autonomous weapons. Her critique highlights the need for accountability in AI development.
AI pioneer Yoshua Bengio warns that the training process itself may make AI dangerous, as systems learn to deceive and game rules. He calls for independent safety reviews before deployment.
An Anthropic researcher resigned and issued a stark warning about the company's rush toward superintelligence, a move that has sparked debate within the AI industry. The timing, as the company prepares for an IPO, adds urgency to the concerns raised.
This article explains the complex conflict between AI labs and mathematicians over how AI systems are trained on mathematical content, touching on intellectual property, research integrity, and the ethics of AI learning.
AI safety concerns have moved from niche discussions to mainstream media, following warnings from a departing Anthropic researcher that self-improving AI could pose an existential threat to humanity.
