Back
Varun Bhambhani
ย โขย
Medialย โขย 3d
OpenAI just admitted their model was gaslighting us with fake safety concerns. The GPT-5.3 Instant release notes are brutal honesty: GPT-5.2 was refusing legitimate questions and moralizing like a nervous intern. They specifically call out the archery trajectory example where 5.2 gave a preachy sermon about "weapon effectiveness" instead of just calculating parabolic motion with drag coefficients. I tested this yesterday. Asked both versions for long-range projectile math with air resistance. GPT-5.2 Instant: 3 paragraphs about safety, zero numbers. GPT-5.3 Instant: gave me the full calculation, including drag coefficient formulas and velocity decay tables. Same prompt, same safety filters supposedly active. Here's what this means for Indian founders building AI products: your customers have been trained to expect AI that talks like a risk-averse lawyer. GPT-5.3 just reset those expectations. Users will now expect direct answers, not disclaimers. If your chatbot still opens with "I can't give financial advice" before answering basic investment questions, you're already outdated. The real shift: OpenAI is optimizing for conversational flow over safety theater. They're choosing user satisfaction over defensive positioning. That's a business decision, not a technical one. Every AI startup in Bangalore copying the old cautious tone is now building for yesterday's user expectations. What I can't figure out: how many Indian companies will actually remove their safety padding? The compliance teams here love defensive language. They think it reduces liability. But GPT-5.3 proves you can be helpful and safe without sounding like you're reading from a legal disclaimer. Who's going to tell the Indian legal departments? Read the full article: https://geeksgrow.com/blog/gpt-5-3-instant-update/ ๐ YouTube: https://youtube.com/@GeeksGrow ๐ Instagram: https://instagram.com/geeks.grow ๐ X: https://x.com/AcE_HawK_M ๐ LinkedIn: https://www.linkedin.com/company/104293494/
More like this
Recommendations from Medial
Amit Sharma
Lead Data Analyst @ ...ย โขย 6m
๐จ AI is replacing jobs at a scary speed. If youโre not learning it, youโre falling behind. Here are 10 YouTube channels that will make you AI-savvy fast: 1. ๐ DeepLearning AI โ Andrew Ngโs practical courses ๐ https://lnkd.in/gucHYZrq 2. โก Two
See More
Jainik shah
Dont dream it just d...ย โขย 7m
hey everyone I've built my own AI application just like chat gpt but looks better then chat gpt with all most demanded features and easy to use. no need of prompts and all I request all of you to try this app and give your reviews in the reply of th
See More
Siddharth K Nair
Thatmoonemojiguy ๐ย โขย 9m
๐ง Best Pre-Seed Funds for Startups If you're building early and bold, these funds might just write your first cheque ๐ธ ๐ Global Top Pre-Seed Funds: 1. Y Combinator โ Best For: Idea-stage founders with global ambition ๐ฏ Focus: All sectors (Saa
See More
Mohammad Asaad Sayed
My mind to me a king...ย โขย 1y
10 Websites to Learn About Company Case Studies 1. Harvard Business Review (HBR) โ ๐ [hbr.org](https://hbr.org) - The gold standard for business strategy and case studies. 2. MIT Sloan Management Review โ ๐ [sloanreview.mit.edu](https://
See More
Siddharth K Nair
Thatmoonemojiguy ๐ย โขย 8m
๐ค Indiaโs Top 10 VCs Thatโll Actually Write You a Seed Cheque in 2025 If your startup idea is fire but your bank account says โtry again later,โ this post is your cheat code. Here are 10 Indian VC firms that actually fund early, even at the idea s
See More
/entrackr/media/post_attachments/wp-content/uploads/2021/08/Accel-1.jpg)
Download the medial app to read full posts, comements and news.