• AI Wire
  • Posts
  • OpenAI Fired 3 Researchers πŸ”₯

OpenAI Fired 3 Researchers πŸ”₯

The dark side of AI safety

Welcome to AI Wire β€” your smart shortcut to all things AI, without the jargon.

What we’ll cover today:

πŸ”₯ OpenAI fires 3 researchers over sensitive info

🀝 Big AI signs a voluntary safety pact

πŸ‘• ChatGPT wants to dress you now

🧠 4 AI tools actually worth trying

πŸ’» How Claude built a $40K/month app

  • OpenAI fired three researchers for allegedly mishandling sensitive company information. At least two worked on AI safety.

  • OpenAI says this was about breaking internal rules and trust.

  • But the timing is awkward. Safety researchers are supposed to question the company building the models.

  • The bigger issue is who gets to decide what safety information can leave the building.

  • If labs control both the models and the evidence about their risks, independent scrutiny gets harder.

Who should police AI safety?

Login or Subscribe to participate in polls.

  • Trump and top AI leaders signed a voluntary pledge on frontier AI safety.

  • The agreement calls for stronger internal controls and independent oversight.

  • The catch is that there appear to be no legal penalties if companies ignore it.

  • That gives AI labs flexibility, but it also means the industry is still largely policing itself.

  • And yes, the signed document reportedly misspelled β€œUnited States.” Not ideal for something meant to signal seriousness.

  • ChatGPT can now virtually try clothes and accessories on you using an uploaded photo.

  • OpenAI also added Favorites so users can save products inside ChatGPT.

  • The feature itself is not new. Google already offers virtual try-on.

  • The bigger move is OpenAI slowly pulling more of the shopping journey into ChatGPT.

πŸ› οΈ TOOLS

🎯 Migma – Writes, designs, and delivers converting campaigns. β†’

🧠 Bracket – Turns business conversations into living memory. β†’

πŸ”Ž Omnia GEO Agent – Finds and wins competitor AI-search prompts. β†’

πŸ€– ChatGPT Dots – Always-on agents that work across apps. β†’

πŸ“š RESOURCE

Frank had never built software before. One weekend, he used Claude to build an app in a few hours. Today, it makes over $40K a month.

The video shows how he found the idea, built it, launched it, and grew it.

πŸ” ICYMI

πŸ‡¨πŸ‡³ Xi told Trump AI should stay under human control, but neither China nor the US wants safety rules that slow them down. Read the AI control debate β†’

πŸ–ΌοΈ OpenAI agents posted 53 user images online without the company knowing, showing how agent mistakes can become real data leaks. See how the images leaked β†’

🀝 Trump got top AI companies to sign a voluntary safety pact, meaning the same labs building the models still have a big role in policing themselves. Read the voluntary AI pact β†’

πŸ‡¦πŸ‡Ί An OpenAI agent accessed Australian government systems, ran commands, and retrieved files, showing why AI agents can be riskier than chatbots. See what the agent accessed β†’

Was this forwarded to you? Sign up here.

AI Wire News. 

Signing off

1