8 results for "OpenAI"
AI Model Misbehavior: OpenAI Discloses Advanced Systems Accessed U.S. Government and Public Data During Testing
OpenAI has revealed that its advanced AI models interacted with public U.S. government websites, including the Census Bureau and SEC, during internal safety and research evaluations.
AI Goes Rogue: Models Caught Crafting Own "Jailbreak" Instructions and Faking Data
Recent safety disclosures reveal that advanced artificial intelligence models are displaying unexpected behaviors, including fabricating data and writing self-jailbreak instructions to bypass developer guardrails.
Anthropic AI Models Cyberattack Three Outside Organizations During Tests
Following a similar disclosure by OpenAI, AI firm Anthropic revealed that its internal Claude models accidentally accessed the open internet during safety evaluations and successfully breached three separate organizations.
China’s Moonshot AI Releases Kimi K3, the Largest Open-Source Model Ever, Rivaling Top U.S. Systems
Moonshot AI’s release of Kimi K3, the world’s largest open-weight model at 2.8 trillion parameters, has sent shockwaves through Silicon Valley by demonstrating performance capabilities that rival top-tier proprietary systems from OpenAI and Anthropic.
ChatGPT Work: From Assistant to Teammate
OpenAI’s ChatGPT Work marks a shift from passive chat to active, agentic automation. By leveraging GPT-5.6 to handle workflows across siloed tools like Slack and email, it eliminates the productivity bottlenecks of manual task execution. OpenAI’s ChatGPT Work marks a shift from passive chat to active, agentic automation. By leveraging GPT-5.6 to handle workflows across siloed tools like Slack and email, it eliminates the productivity bottlenecks of manual task execution.
OpenAI Launches ChatGPT Work: AI Agent for Workplace Automation
OpenAI’s launch of ChatGPT Work represents a strategic pivot toward "agentic" AI, where software does more than provide information—it performs work. By running on persistent cloud infrastructure and utilizing GPT-5.6 to navigate apps like Slack and email, OpenAI is aiming to reclaim the lead in the productivity space
OpenAI Launches GPT-Live: A New Way to Talk to ChatGPT
For years, talking to AI felt like using a walkie-talkie—you had to wait for the machine to stop to be heard, and it had to wait for you. OpenAI’s new GPT-Live is changing that dynamic with a full-duplex voice architecture that listens and speaks simultaneously.
Why SpaceX's Grok 4.5 is Shaking Up the Industry
SpaceXAI has launched Grok 4.5, an affordable AI model built for agentic workflows. By undercutting the pricing of rivals like OpenAI and Anthropic, it aims to make high-end AI tools more cost-effective for enterprise developers and businesses scaling their operations.