Hacktron exploits OpenAI forum flaw with Claude Opus 5 in 3 hours
The story in brief
Security researchers from Hacktron discovered a flaw in OpenAI’s public help forum, allowing them to take control. They used Anthropic’s Claude Opus 5 to exploit the vulnerability within three hours. OpenAI fixed the issue in 14 hours and paid a $6,500 reward. The flaw is not unique to OpenAI and affects products from Slack and Meta.