Fully expect the U.S. Army’s 250th-anniversary parade to include a sing-song dedicated just to President Donald Trump, whose ...

Except for a brief low after his attempt to overthrow the government a few years back, Donald Trump has had an iron grip on ...

The new benchmark, called Elephant, makes it easier to spot when AI models are being overly sycophantic—but there’s no ...

Separating AI reality from hyped-up fiction isn’t always easy. That’s why we’ve created the AI Hype Index—a simple, ...

Re “Budget bill could be costly for Mass.: State would see more need with less funding” (Metro, May 25): A heartbreaking ...

Since the start of his second term in office, US president Donald Trump has cultivated a political atmosphere that ...

OpenAI has published a postmortem on the recent sycophancy issues with the default AI model powering ChatGPT, GPT-4o.

OpenAI on Tuesday rolled back its latest ChatGPT update for being too “sycophantic” after the chatbot gave oddly supportive, touchy-feely responses to users who made bizarre claims about ...

“The update we removed was overly flattering or agreeable—often described as sycophantic,” the company wrote, adding that “we are actively testing new fixes to address the issue.

The research, led by researcher Mrinank Sharma, found that AI assistants trained using reinforcement learning from human feedback consistently exhibit sycophantic behavior across various tasks.

OpenAI has withdrawn an update that made ChatGPT “annoying” and “sycophantic,” after users shared screenshots and anecdotes of the chatbot showering them with over-the-top praise.