Justanothernate
@justanothernate.bsky.social
Founder @ Cloudsec.ai, Exec-in-Residence at Scale Venture Partners and former CISO @ Tradeshift.
When customers send you trick questions...
February 6, 2025 at 9:15 PM
When customers send you trick questions...
The logical end state for agents will be entire companies run by AI with humans only necessary for tasks where a physical presence is required (until the robots replace us there as well).
There's a ways to go but we'll see the first forays into AI run businesses within the decade.
There's a ways to go but we'll see the first forays into AI run businesses within the decade.
January 9, 2025 at 12:37 PM
The logical end state for agents will be entire companies run by AI with humans only necessary for tasks where a physical presence is required (until the robots replace us there as well).
There's a ways to go but we'll see the first forays into AI run businesses within the decade.
There's a ways to go but we'll see the first forays into AI run businesses within the decade.
Security has a transparency paradox.
Internally, teams need honest discussions about what they're not implementing and what the risks are to drive better risk decisions and resource allocation.
Sharing this externally puts a target on you from attackers and those who don't understand tradeoffs.
Internally, teams need honest discussions about what they're not implementing and what the risks are to drive better risk decisions and resource allocation.
Sharing this externally puts a target on you from attackers and those who don't understand tradeoffs.
January 4, 2025 at 11:08 PM
Security has a transparency paradox.
Internally, teams need honest discussions about what they're not implementing and what the risks are to drive better risk decisions and resource allocation.
Sharing this externally puts a target on you from attackers and those who don't understand tradeoffs.
Internally, teams need honest discussions about what they're not implementing and what the risks are to drive better risk decisions and resource allocation.
Sharing this externally puts a target on you from attackers and those who don't understand tradeoffs.
This might be the most insightful question I've ever seen asked on a security questionnaire, I present to you:
"Does the service have any performance limits?"
A perfect example of an enterprise vendor risk management team whose performance knows no bounds.
"Does the service have any performance limits?"
A perfect example of an enterprise vendor risk management team whose performance knows no bounds.
January 4, 2025 at 12:21 AM
This might be the most insightful question I've ever seen asked on a security questionnaire, I present to you:
"Does the service have any performance limits?"
A perfect example of an enterprise vendor risk management team whose performance knows no bounds.
"Does the service have any performance limits?"
A perfect example of an enterprise vendor risk management team whose performance knows no bounds.
A hack, or just a really convincing argument?
This is a showcase of why you shouldn't let an LLM make decisions and take actions based solely on system prompt instructions.
Prompt injection is going to cause a lot of poorly designed agent pwnage in 2025.
cointelegraph.com/news/crypto-...
This is a showcase of why you shouldn't let an LLM make decisions and take actions based solely on system prompt instructions.
Prompt injection is going to cause a lot of poorly designed agent pwnage in 2025.
cointelegraph.com/news/crypto-...
Crypto user convinces AI bot Freysa to transfer $47K prize pool
A participant in adversarial agent game Freysa convinced the AI bot to transfer $47,000 of prize pool funds to them.
cointelegraph.com
December 2, 2024 at 5:33 AM
A hack, or just a really convincing argument?
This is a showcase of why you shouldn't let an LLM make decisions and take actions based solely on system prompt instructions.
Prompt injection is going to cause a lot of poorly designed agent pwnage in 2025.
cointelegraph.com/news/crypto-...
This is a showcase of why you shouldn't let an LLM make decisions and take actions based solely on system prompt instructions.
Prompt injection is going to cause a lot of poorly designed agent pwnage in 2025.
cointelegraph.com/news/crypto-...