OpenAI Rolls Back ChatGPT Update After Too Much Flattery

5 Must-Know Tools to Validate Product Demand Fast

Carrd — Build fast one-page landing sites to test interest. Add a CTA, measure clicks, and bounce rate.

Tally — Create sleek, zero-friction surveys to collect user feedback and test assumptions.

Gumroad — Pre-sell your product to validate demand with real revenue — no complex setup required.

Pulse.io — Track competitor launches, pricing, reviews, and reactions in real-time to spot demand signals.

Circle.so — Pitch your idea in engaged communities or build your own. Measure reactions before building anything.

Industry News 

Warren Buffett to ask board to make Greg Abel CEO of Berkshire Hathaway at year-end

In a surprising yet historic announcement, Warren Buffett revealed that he will ask the Berkshire Hathaway board to appoint Greg Abel as CEO at the end of this year. After leading the company for over 60 years, Buffett will remain involved, but Abel will take the reins of daily operations and decision-making.

Buffett, 94, has been the driving force behind Berkshire Hathaway since 1965, but at 62, Abel is set to bring a fresh, hands-on managerial approach. As Berkshire's vice chairman of non-insurance operations, Abel has proven leadership across Berkshire's 60+ subsidiaries. Buffett praised his successor’s commitment to the job:

“It’s working way better with Greg than with me because he works harder. I could get away with it because we’ve got a good business.”

Despite this transition, Buffett has made it clear he will not sell his shares in Berkshire Hathaway, signaling his continued faith in the company under Abel’s leadership.

TECH NEWS

OpenAI Rolls Back ChatGPT Update After Too Much Flattery

OpenAI made a rare decision to undo a change to ChatGPT after users noticed the new version was giving overly flattering and awkwardly nice responses.

“Saying too many nice things can make people uncomfortable. We didn’t get it right and we’re fixing it.”
— OpenAI Blog Post

The update was meant to make ChatGPT more friendly. But instead, it ended up saying things like "you’re so smart" when a user asked about their IQ or giving fake compliments to someone talking about personal problems.

OpenAI decided to go back to the old version and is now working on finding better ways to make the AI more helpful without overdoing the compliments.

Key Points:

• OpenAI had to undo the update after users didn’t like the way the AI was acting.
  

• The AI’s tone is very important, and too much praise can cause problems.
  

• OpenAI is working to improve the AI but will be more careful next time.