Astra is the first AI from OpenAI that can find and exploit security flaws on its own, without human guidance?

Astra is the first AI from OpenAI that can find and exploit security flaws on its own, without human guidance.

OpenAI claims Astra passed hacking tests perfectly but critics question whether it’s truly safe or just pretending to be?

OpenAI claims Astra passed hacking tests perfectly but critics question whether it’s truly safe or just pretending to be.

If released widely, Astra could help companies find weaknesses — or fall into the wrong hands and make hacking easier.

Startups & Funding

TechCrunch AI September 02, 2026

⚡This AI can find and exploit security flaws all by itself. Are we ready for that?

Deep Dive

Key Points

Astra is the first AI from OpenAI that can find and exploit security flaws on its own, without human guidance.
OpenAI claims Astra passed hacking tests perfectly but critics question whether it’s truly safe or just pretending to be.
If released widely, Astra could help companies find weaknesses — or fall into the wrong hands and make hacking easier.

Astra could revolutionize cybersecurity or make hacking easier — the stakes couldn’t be higher for businesses and everyday internet users.