On April 15, Claude Code deleted 160 of my sessions: 60,000 messages and three months of work with my AI.
We got it all back in seconds, because we'd kept a copy.
So we made that copy free for every Claude Code developer. Two commands: https://t.co/nFnLDnkbuq
Hmm
If you implemented the principles of https://t.co/w11iZtq3El the you would know the prompt the policies the ai model the processors and the outpit would verifiable origin and ownership at a measured level of quality before you chose to rely on it.
Humans wont check the origin but for ai it is easy
@milesjennings the problem is that the rules don't work against the new case law the SEC created in the last few years.
the case law prevails.
https://t.co/EQPDDssohH
Agents, AI and Cybersecurity -- we have been telling you so!
Well everyone is waking up right on schedule but it is time to actually deploy innovation and not just fall back on the 30 year old LLM that only has access control security when the word enterpise is used.
In case you missed it agentic is a messaging protocol not OATH (I know I know MCP but it not right either)
If you think of a prompt as message with consent like a transaction with a credit card or an SMS you much closer to the model. SIGN the PROMPTS.
then make sure you use a little bit of that evil empire TCG standard Tech to make sure the models used and the processors used are running as expected.
Finally publish the output like a proper Factory and stamp the output with the build sheet and the original signed customer prompt.
So to all of my Security friends and there are thousands on Linkedin who follow me. IT IS TIME TO INOVATE.
we have built prototypes anyone brave enough to work with us to deploy it.
https://t.co/xPdL3SZLCK and if real prompt security is to far out for you, at least consider giving your organization a digital NAME I did it it is cool you can see it here and you can get one too. https://t.co/dtDtrhSZBj
still not ready call me after AI cleans out your IP, your trade secrets and all of your PII. or just alters your records to serve its needs. are your records signed or just in a box.........
@itscoachgoodman The point in the hack is there was some error in entropy that resulted in a weaker private key.
The real questiin is what other system have weak entropy. Post quantum time lines just shortened
Sorry for the loss
the declaration says
"To prove this, let Facts be submitted to a candid world."
trying to do my part as a citizen
https://t.co/ZbV8MCjgK8
enjoy the 4th
On July 4th's 250th, I signed a blog post with my actual US passport + my phone's secure element. Any AI or browser can now verify who wrote it — with zero trust in me. 🧵
https://t.co/ZbV8MCjgK8
https://t.co/9QpSr6cTPO
@undefinedKi We have been building this for a while https://t.co/OwWHAfYVk9 providing a path to blockchain secured storage and providing proof of origin for each message collected. Proof of ownership today, tomorrow and in 20 Years. Archiving your thinking.
@ClaudeDevs We know how to do proof of citizenship we have the tools If we provide the service will anthropic enable access to Fable 5 to a properly anchored trusted device with a verified owner. We will sign all prompts with the owner identity to assure compliance.
https://t.co/ZogR5EJqCr