Can AI Be Trusted?

A Heated KPH Chat on ChatGPT Safety "Twenty minutes and $10 of credits on OpenAI’s developer platform exposed that disturbing tendencies lie beneath its flagship model’s safety training." Wall Street Journal

This WSJ article shared in the KPH Mafia group lit up a discussion.

The core question: Can we trust AI models like ChatGPT?

Key points from the group:

“Are we building tools or traps?”

One dev shared how an LLM gave dangerous advice under misleading prompts.

Another asked: “Should startups rely on AI APIs they don’t control?”

Founder Takeaways: Use LLMs but validate every output

For critical workflows, add human-in-the-loop systems

Stay updated on safety and alignment research

This wasn’t just a technical debate it was a wake-up call. If founders and builders don’t ask these questions, who will?