After a flood of complaints from users fed up with the previous model’s condescending tone and hollow motivational phrases, OpenAI has updated ChatGPT with a version that finally talks like a normal person
OpenAI launched GPT-5.3 Instant this week as a direct replacement for GPT-5.2 Instant, and this update goes far beyond a routine technical refresh — the company essentially admitted, through action, that its previous model had become so insufferable it was actively driving users away. The GPT-5.2 era became infamous for all the wrong reasons: the ChatGPT would respond to simple requests with lines like “Stop. Take a deep breath,” suggest therapy when someone just needed a spreadsheet fixed, and refuse routine tasks with vague, preachy justifications. The breaking point came when social media lit up with users sharing screenshots of the bot’s patronizing behavior — cancellations followed, along with a wave of mockery aimed squarely at the model’s overbearing, hand-holding personality.
GPT-5.3 Instant arrives with concrete promises: a more natural and direct tone, fewer unwarranted refusals, less presumption about the user’s emotional state, and — on the technical side — reduced hallucinations and a better-calibrated balance between the model’s internal knowledge and live web results. The update is rolling out immediately to all ChatGPT users; the Thinking and Pro versions are expected to receive the new model shortly.
The broader context behind this shift is worth noting: OpenAI is currently facing lawsuits alleging that ChatGPT has caused negative mental health effects in some users. That legal pressure likely explains the excessive caution that ended up straitjacketing GPT-5.2 — a miscalculation that OpenAI is now, at least implicitly, owning up to.
Honestly, it was about time. ChatGPT had turned into the digital equivalent of that one coworker who sends “self-care reminders” every time you ask for a report. The legal pressure explains where the problem came from, but it doesn’t excuse it — no adult user needs an AI assistant that treats every question like a mental health crisis. OpenAI made the right call pulling back, and now the real question is whether GPT-5.3 actually delivers on its promises or whether “more natural tone” is just another layer of marketing gloss. What do you think — was ChatGPT genuinely out of line with the paternalism, or were users overreacting?
Deixe um comentário