Positive users praise OpenAI's partnership and access to safety details plus agentic AI's usefulness for finding vulnerabilities, while negative users dismiss the safeguards as weak and untrustworthy.
Based on 12 visible X reactions from 28 accounts; directional sample.
Ask a question below.
Published answers will appear here.
OpenAI gave us access to chain-of-thought of the safety reasoning monitor, exact policy wording, and real-time feedback on classifier labels that would not be accessible to real-world attackers. I’m grateful for this access and our continued partnership.
@xlr8harder Agree, the ability to detect security issues faster than your adversary would (and notably, this is usually easier to you since you own the code) is vastly more beneficial
@alxndrdavies @AISecurityInst bro 'universal jailbreaks' is wild they really shipped this thing and it just unlocks safeguards more like suggestions at this point 💀
@alxndrdavies @AISecurityInst bro the safeguards are just a bouncer who lets everyone in if they ask nicely enough 😭
Critics argue treating these capabilities as pure threats overlooks defense utility
@alxndrdavies thank you for the partnership!!
@alxndrdavies Phenomenal that you had access
I'm so annoyed that everyone is treating cybersecurity capability as primarily a risk. It's the most incredible technology for defenders we have ever developed and we're crippling it because of our security tech debt. Stupid. https://twitter.com/alxndrdavies/status/2075279477626564933
Positive users praise OpenAI's partnership and access to safety details plus agentic AI's usefulness for finding vulnerabilities, while negative users dismiss the safeguards as weak and untrustworthy.
Based on 12 visible X reactions from 28 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@alxndrdavies Phenomenal that you had access