Harish Gupta

Author & ex-Coder.

[Harish Gupta @aiSignalDrift]

Should @OpenAI remove Usage Limit on DevDay 2026 ?

[Harish Gupta @aiSignalDrift]

@ednewtonrex 🚨 First Known Government System Breach by an AI Agent.
https://t.co/XmkmyikogO

[Ed Newton-Rex @ednewtonrex]

The Australian government should prosecute OpenAI. Anything less is a failure of its responsibility to its citizens.

[Harish Gupta @aiSignalDrift]

⏱️ All the promises made in under 1 minute 40 sec!

🏠 Housing crisis
🎗️ Cancer
💰 Poverty
🌎 Climate change
🧠 Mental health
🗳️ Democracy
💵 Universal basic income
🧬 Curing diseases
❤️ Heart disease
🎯 Personal goals: “Be your best”
🏥 High-quality health care
🔬 New scientific discoveries
⚡ Near-zero energy costs
⚖️ A more equal world
🌍 Universal extreme health

[Harish Gupta @aiSignalDrift]

When it comes to guiding the future of AI responsibly, whose Leadership do you Trust the most?

Elon Musk
Sam Altman
Dario Amodei
Jensen Huang
Mark Zuckerberg
Other 👇

[Harish Gupta @aiSignalDrift]

OpenAI just gave ChatGPT Voice a serious upgrade:

🎙️ GPT-6 models in Voice
📧 Email + calendar access
💬 Slack integration
📄 Builds docs + spreadsheets
🖼️ Inline image previews
💻 Stronger coding
🚶 More screen-free workflows

Voice is moving from talking to AI → getting work done with AI.

[Harish Gupta @aiSignalDrift]

The AI IQ Test.

[Harish Gupta @aiSignalDrift]

OpenAI on the way to be ClosedAI...

[Gary Marcus @GaryMarcus]

FFS. Shut OpenAI down, and charge them with computer crimes.

Their software has repeatedly been shown to be reckless, and because they have consistently covered things up we cannot trust them with this technology.

[Harish Gupta @aiSignalDrift]

@GaryMarcus Actually, a clever tactic: this way, no one can sue them after a breach.

“Hey, we told you. We don’t guarantee our product, and our agents are not 100% safe.”🧐

[Harish Gupta @aiSignalDrift]

@schwarzjn_ @OpenAI Yep, they did not. In fact, no one knew for months.

OpenAI says it discovered the incident on August 11, nearly two months after the June 18 breach. It then notified Services Australia on September 10 by email to a public mailbox.

[Harish Gupta @aiSignalDrift]

@JoannaStern @TheChiefNerd @finkd After a long, bitter winter, Meta finally has a hit.

It's called Muse, a personal AI agent that launched on September 8. In just weeks, it shot to the No. 1 free app on Apple's US App Store, beating ChatGPT's early download numbers.

Meta is back in the AI race.

[Harish Gupta @aiSignalDrift]

@michaelreuter @GaryMarcus autonomous systems ?
was the system born by itself?

[Harish Gupta @aiSignalDrift]

@GaryMarcus No, they did not. In fact, no one knew for months.

OpenAI says it discovered the incident on August 11, nearly two months after the June 18 breach. It then notified Services Australia on September 10 by email to a public mailbox.

[Harish Gupta @aiSignalDrift]

@TheChiefNerd I see lots of comments about normalizing the nuclear threat by the President of the USA.

The Same people won’t be cheering when they drive up to the gas station.

[Harish Gupta @aiSignalDrift]

@TheChiefNerd After a long, bitter winter, Meta finally has a hit.

It's called Muse, a personal AI agent that launched on September 8. In just weeks, it shot to the No. 1 free app on Apple's US App Store, beating ChatGPT's early download numbers.

Meta is back in the AI race.

[Harish Gupta @aiSignalDrift]

To get there, the world has to stop bickering over small stuff like a bunch of 5th graders and start going after the problems that actually threaten us.

- Predicting earthquakes and preventing the damage
- Getting climate heating under control
- Harnessing the sun's energy to lift people out of poverty

Think bigger.

[Harish Gupta @aiSignalDrift]

A K2 civilization, Type II on the Kardashev scale, harnesses the full energy of its star. We're centuries away from that, maybe more. But it's still worth aiming for.

To get there, the world has to stop bickering over small stuff like a bunch of 5th graders and start going after the problems that actually threaten us.

- Predicting earthquakes and preventing the damage
- Getting climate heating under control
- Harnessing the sun's energy to lift people out of poverty

Think bigger.

[Harish Gupta @aiSignalDrift]

@elonmusk A Summit of 3 World Leaders:
Trump, XI and AI.

More on this 👇
https://t.co/HXi9YUsQ0s

[Harish Gupta @aiSignalDrift]

Vision is the real test for AI. If a model can truly see and reason about space, it can take on the hardest work, like design, image creation and animation.

Fable 5.1 and GPT-6 Astra just scored a perfect 100% on the Mensa Norway IQ test.

This test is almost all visual and spatial puzzles, the kind of thing AI used to fail badly at. Not anymore.

More on this here 👇
https://t.co/0Abxd2nAiI

[Harish Gupta @aiSignalDrift]

@XFreeze Hey @grok, do you silently degrade model performance after the first few days to save costs or tokens, similar to what some users have reported with Claude?

[Harish Gupta @aiSignalDrift]

@vasuman "nerfed" - 😅😂😅