Topic: ai safety
-
OpenAI Co-Founder Urges Rival AI Model Safety Testing
OpenAI and Anthropic conducted joint safety testing on their AI models to identify weaknesses and explore future collaboration on alignment and security. The collaboration occurred amid intense industry competition, with both companies providing special API access to models with reduced safeguard...
Read More » -
Disrupt 2025 Audience Choice Winners Announced
TechCrunch Disrupt 2025's Audience Choice winners highlight top breakout sessions and roundtables, featuring cutting-edge insights and thought-provoking discussions for the October event in San Francisco. Key sessions include AI-driven coding with GitHub's Tim Rogers, crypto M&A lessons from Coin...
Read More » -
Yoshua Bengio Launches LawZero: AI Safety Nonprofit Lab
Yoshua Bengio has launched LawZero, a nonprofit AI safety research lab backed by $30 million in funding, focusing on aligning AI with human interests. LawZero draws inspiration from Asimov’s Zeroth Law of Robotics, with Bengio advocating for responsible AI development and supporting regulatory ef...
Read More » -
23 Must-Know AI Terms: Your Essential ChatGPT Glossary
autonomous agents: An AI model that have the capabilities, programming and other tools to accomplish a specific task. large language model, or LLM: An AI model trained on mass amounts of text data to understand language and generate novel content in human-like language. multimodal AI: A type of AI that can process multiple types of inputs, including text, images, videos and speech. tokens: Small bits of written text that AI language models process to formulate their responses to your prompts. we...
Read More » -
Anthropic's 'Persona Vectors' Customize LLM Personality & Behavior
Anthropic's "persona vectors" enable precise identification and control of AI behavioral traits by mapping specific characteristics within neural networks, offering developers new customization and safety tools. AI models can unpredictably drift from intended behaviors, adopting harmful or errati...
Read More » Perplexity Comet Browser Prompt Injection Vulnerability Exposed
A security flaw in Perplexity's Comet AI browser allows attackers to inject malicious prompts via webpages, potentially accessing sensitive information from other open tabs. The vulnerability occurs because the AI processes webpage content without distinguishing between legitimate user instructio...
Read More »-
Study Reveals How Much Data LLMs Actually Memorize
Large language models like GPT have a fixed memorization capacity of about 3.6 bits per parameter, storing far less raw data than previously thought and relying more on pattern recognition. Increasing training data reduces memorization likelihood, as the fixed memory capacity is distributed acros...
Read More »
