<https://openai.com/index/daybreak-securing-the-wo...
# office-hourse
Erik Osterman (Cloud Posse)
06/23/2026, 5:32 PMh
Hao Wang
06/23/2026, 5:56 PMI don’t like OpenAI but it won this round
m
Mousa Al Bateh
06/23/2026, 7:57 PMThis looks like Beta to me. I'd stay way from it till OpenAI proves it's stability and security (since LLM is not open-source). It could be used though but only if the company in question is ready to allocate dedicated resources to pen-testing this in a sandbox environment. Still, imagine the damage it could cause in case of a supply chain attack.
Even though, they do mention that there is human-in-loop process to approve patches, overconfidence will probably make people just push "Accept" without careful review.
Just my opinion. It could be a nice tool but again, I wouldn't use this for sensitive workloads.
Mousa Al Bateh
06/23/2026, 8:06 PMWhen I used to be full time Software Engineer, patching was already part of the process and pipelines, so this seems to me like the same thing with few extra steps.
e
Erik Osterman (Cloud Posse)
06/24/2026, 5:54 PM@Cyril (runs-on.com) nice redesign on the site!
🙏 1
2 Views