Security researchers claim OpenAI breach via chained flaws, heightening AI-safety fears

A small cybersecurity firm Hacktron says it breached OpenAI earlier this year by chaining two unknown vulnerabilities—one in the third‑party Discourse platform and one in how OpenAI validates employees—giving access to ChatGPT accounts. OpenAI patched the flaws and paid Hacktron a $6,500 bug bounty; no harm was done. The disclosure comes as AI safety concerns surge and regulators and experts warn about security risks in AI ecosystems.