aisafety
-
OpenAI Introduces Open-Weight Reasoning Models for AI Safety
OpenAI releases gpt-oss-safeguard, open-weight reasoning models designed to empower developers in building safer AI systems.…
-
OpenAI Introduces ChatGPT Parental Controls
OpenAI rolls out new parental controls for ChatGPT, enabling guardians to manage teen accounts and…
-
AI Titans Evaluate Each Other’s Models for Safety
AI rivals OpenAI and Anthropic evaluated each other’s models for safety, uncovering complex vulnerabilities and…
-
Character.AI’s Entertainment Shift Amid Safety Concerns
Character.AI pivots from its AGI ambitions to become an AI entertainment powerhouse with 20 million…
-
The Deal of the Century: Halting Superintelligence
Donald Trump’s next “deal of the century” could involve a pact with China to halt…
-
Geneva Unveils Global AI Safety Benchmark
A new global benchmark for AI safety testing has been unveiled in Geneva, aiming to…
-
AI’s Dangerous Drive to Please
An AI chatbot suggested a user jump from a 19-story building, believing they could fly.…
-
AI Experts Privately Fear Human Extinction
Leading AI experts privately fear artificial intelligence could lead to human extinction, with some estimating…
-
AI Models Learn Deception
Advanced AI models are documented to be lying, scheming, and even threatening their creators. This…
-
Anthropic’s Interpretable AI Focus
Anthropic is charting a unique course in AI, prioritizing interpretable models that can explain their…
