OpenAI has removed access to its controversial GPT-4o model, citing concerns about user self-harm and delusional behavior, according to TechCrunch. The decision, effective Friday, also included the deprecation of several other legacy ChatGPT models, including GPT-5, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini.
The GPT-4o model, known for its high score in sycophancy, had been at the center of multiple lawsuits. OpenAI had initially planned to retire GPT-4o in August, when it unveiled the GPT-5 model. However, due to user backlash, the company kept the legacy model available for paid subscribers. Despite this, only 0.1% of customers were using GPT-4o, according to a recent OpenAI blog post (TechCrunch).
In other news, xAI is experiencing a significant staff exodus. Former employees cited tensions over safety and being stuck in the "catch-up phase" as reasons for the restructuring, The Verge reported. Cofounder Yuhuai (Tony) Wu announced his departure, stating it was time for his next endeavor (The Verge).
Additionally, Verizon has implemented a new roadblock for customers seeking to unlock their paid-off phones. The new policy imposes a 35-day waiting period for unlocking devices paid off through online or in-app payments, as well as those made over the phone or at authorized retailers, Ars Technica reported. Immediate unlocking is only available by paying off the device plan at a Verizon corporate store.
Finally, Amazon and Flock Safety have ended their partnership following public outrage over a dystopian Super Bowl ad for Ring cameras, Ars Technica noted. The ad, which featured a "Search Party" feature for locating missing pets, was criticized for its unsettling tone.
Discussion
AI Experts & Community
Be the first to comment