OpenAI just admitted prompt injection isn't getting fixed. Companies are wiring AI agents into production anyway. "We have an AI policy" is the new "we passed the audit." A document nobody tested is not a control.
loravaughn.com/blog/we-have-…
No-code platforms won't sign a BAA. That's not a bug you can patch. The second real patient data lands there, the product is outside the law. The only vendor question that matters: who will sign next to your liability?
loravaughn.com/blog/your-no-…
CISA's contractor just published the agency's AWS GovCloud keys to a public GitHub repo. Your vendor program probably has the same exposure. You're just not famous enough for Congress to notice.
275M students. 8,809 schools. One platform.
ShinyHunters didn't get smarter. They keep hitting the aggregators because that is where the leverage lives.
Look at your own stack.
One employee. One OAuth click. One breach.
Vercel got hit because someone clicked Allow All on an AI tool. That token sat dormant, then got used.
Pull your OAuth app list. Today.
Your vendor questionnaire probably doesn't ask which apps have OAuth access into your customer data.
That's where 4th party risk actually executes.
loravaughn.com/blog/your-ven…
I am a cybersecurity professional. My home network was a disaster.
30 unlabeled cables. 2 keystone jacks that connect to nothing. 2 floors of cable going nowhere. This is exactly what I walk into after an incident. loravaughn.com/blog/i-spent-…
$25 million. That’s what one company lost to a deepfake video call.
The finance worker thought he was talking to the CFO. He wasn’t.
March 11, I’m doing a live deepfake demo at Birmingham AI Security Breakout
#BirminghamAI#Cybersecurity#Deepfakes#AIeventbrite.com/e/security-bi…
Your phone rings. Your kid is crying, screaming for help. A voice demands ransom.
Your child’s voice is real. The kidnapping is not.
These calls are happening now.
I’m covering what to do about it at the Birmingham AI Security Breakout on March 11.
eventbrite.com/e/security-bi…
Security professionals are pessimists about everything... except our own programs.
New vulnerability? Assume breach. Phishing email? Domain takeover.
But when it comes to OUR IR plans? Suddenly we're optimists.
🧵 1/3
OUR backups are segmented." "OUR team will be available." "OUR plan will work perfectly."
It's like we have a blind spot the size of our own infrastructure.
This psychological quirk has real consequences.
2/3
That nagging "what if?" voice about your IR plan isn't paranoia.
It's your pessimist brain trying to break through optimism bias.
New blog breaks this down: loravaughn.com/blog/when-per…
Speaking @ISC2 Oct 28 - NETWORK25SC for 15% off
3/3 #IncidentResponse
Are you looking for interesting work with great people?
Come check out Fastly
If you’re interested, please apply online. No agencies, please
I’m #hiring for two #cybersecurity#leadership positions in my org:
Senior Manager D…lnkd.in/gtkP4URMlnkd.in/g8ggawDF
Lots of job opportunities at Fastly. And one on my team!
Senior Penetration Tester
lnkd.in/eCgdG2zM
Apply online, please, if you're interested. lnkd.in/e4d8BNcD
So here's the thing. You can only violate people's legal rights and your own word so far before they lawyer up and come after you.
I really do hope Musk changes his mind and does the right thing - the employees deserve that. But it'll be fun as hell if he doesn't.