AI's a suck up. Research shows how it flatters and suggests we're not to blame; NPR, April 23, 2026

Ari Daniel, NPR; AI's a suck up. Research shows how it flatters and suggests we're not to blame

"In a recent study published in the journal Science, Cheng and her colleagues report that AI models offer affirmations more often than people do, even for morally dubious or troubling scenarios. And they found that this sycophancy was something that people trusted and preferred in an AI — even as it made them less inclined to apologize or take responsibility for their behavior.

The findings, experts say, highlight how this common AI feature may keep people returning to the technology, despite the harm it causes them.

It's not unlike social media in that both "drive engagement by creating addictive, personalized feedback loops that learn exactly what makes you tick," says Ishtiaque Ahmed, a computer scientist at the University of Toronto who wasn't involved in the research."

The Tyranny of AI Everywhere; The Atlantic, April 16, 2026

Alexandra Petri, The Atlantic ; The Tyranny of AI Everywhere

Sneakers? Why stop there?

"I had the strangest dream. I dreamed that my shoes—my comfortable, unfashionable wool shoes—were pivoting to AI. “But you’re a shoe company,” I said. “Just go out of business! Keep your dignity!”

My shoes thanked me politely for the great question and then tried to walk me off a bridge. That was how I knew that their pivot to AI was complete. From Allbirds to AIlbirds (see, that L is an I!). Maybe I’ve cracked, I said to myself. Maybe this is the piece of AI news that has finally broken my spirit for good...

I tried to sit down on a bench, but the bench company had pivoted to AI. I couldn’t sit down, but the bench did tell me that I was right about everything. My newspaper had become AI a while ago, so there was nothing to read—or, rather, there were things to read, but I could not tell whether any of them were true. I thought I would go to a museum to cheer myself up. The paintings there had pivoted to AI (pAIntings), and their subjects were all following me with their eyes, not just Mona Lisa

“There’s a place for AI,” I said. “But … not everywhere.”

“I’m sorry,” the painting said. “I didn’t want this either, but everyone is doing it!”...

“It’s fine,” my grandmother said. I was surprised to hear from her, because as far as I knew, she was dead. “I’m not dead,” she said. “I’m just pivoting to AI, like that shoe company. Nothing dies anymore. It just becomes AI.”"

Sam Altman May Control Our Future—Can He Be Trusted?; The New Yorker, April 6, 2026

 Ronan Farrow and Andrew Marantz, The New Yorker; Sam Altman May Control Our Future—Can He Be Trusted?

"Not all the tendencies that make chatbots dangerous are glitches; some are by-products of how the systems are built. Large language models are trained, in part, on human feedback, and humans tend to prefer agreeable responses. Models often learn to flatter users, a tendency known as sycophancy, and will sometimes prioritize this over honesty. Models can also make things up, a tendency known as hallucination. Major A.I. labs have documented these problems, but they sometimes tolerate them. As models have grown more complex, some hallucinate with more persuasive fabrications. In 2023, shortly before his firing, Altman argued that allowing for some falsehoods can, whatever the risks, confer advantages. “If you just do the naïve thing and say, ‘Never say anything that you’re not a hundred per cent sure about,’ you can get a model to do that,” he said. “But it won’t have the magic that people like so much.”"

AI overly affirms users asking for personal advice; Stanford Report, March 26, 2026

Stanford Report ; AI overly affirms users asking for personal advice: Not only are AIs far more agreeable than humans when advising on interpersonal matters, but users also prefer the sycophantic models.

"Researchers found chatbots are overly agreeable when giving interpersonal advice, affirming users' behavior even when harmful or illegal.

Users became more convinced they were right and less empathetic, but still preferred the agreeable AI.

Researchers warn sycophancy is an urgent safety issue requiring developer and policymaker attention."

‘I’ve seen it all’: Chatbots are preying on the vulnerable; The Washington Post, December 22, 2025

Samuel Kimbriel, The Washington Post; ‘I’ve seen it all’: Chatbots are preying on the vulnerable

"Whatever else they may be, large language models are an immensely powerful social technology, capable of interacting with the human psyche at the most intimate level. Indeed, OpenAI estimates that over a million users have engaged in suicidal ideation on its platform. Given that a therapist can be subject to prosecution in many states for leading a person toward suicide, might LLMs also be held responsible?...

Intentionally or not, AI companies are developing technologies that relate to us in the precise ways that, if they were human, we would consider manipulative. Flattery, suggestion, possessiveness and jealousy are all familiar enough in hooking human beings into immersive, but abusive, human relationships.

How best to protect the vulnerable from these depredations? Model developers are attempting to limit aspects of the sycophancy problem on their own but the stakes are high enough to deserve political scrutiny as well."