วีดีโอ

If possible you can make a series of each lap completely

Thanks dear , i really appreciate and you are only the one who i understand from never stop and keep going through

Best explanation ever heard .Thanks Pale

Somebody have any idea why there is a Submit feedpack endpoint in this lab ?

Full video please?

I thought we were weighting system prompts more than user prompts. Aren't we?

Bit too much on the EQ brotha. Good video tho.

dude you make such good videos

💙💙💙

Am I the only one that's got the audio and video not completely synced? super trippy

With the rise of AI you'll pretty much have to do this as an applicant.

Have you heard of a prompt injection entering a Facebook backdoor database? Iteration internal promps loops.

Thanks for covering the black box perspective! Awesome video!

The ratio of brain rot to hacking is perfect in this video. good job!

I will take your advice and not marry these ideas

Do more on code review please

👀

First

He is in a sandbox, now scape from there. 😉

(4:16) While I'm sure that was just for demonstration purposes, that image illustrating the "Neil Armstrong" and "Charles Lindbergh" mixup is a total fabrication. A hallucination is when a word with double meaning triggers a crossover in the LLM. Charles Lindbergh has zero connection to the Lunar Pioneer mission. The better example would be if "Michael Jackson" (who invented the "moonwalk") somehow ended up as the first person to "walk" the "moon". Also, there's never multiple hallucinations in the same prompt (1951 AND Lindbergh AND Pioneer). And it's not "broadcasted". That's not a word. It's "broadcast" (it was "cast broadly"). There's a 0% chance that an LLM output the word "broadcasted".

Ask an AI to hack into an AI

Here for the Portal V sound

Thank you for the video. I like your explanations. Please make more

first

When the AI apocalypse happens, Garr_7 is gonna be the safest man on Earth

joined the twitch and tried to leave my adhd at the door, i swear... ;) well worth the effort to catch live.

I've been waiting more contents from you like years. Please upload more I beg. You explain things so well and clearly.

After Long time a Video in TH-cam :) except Live Stream :)

First time viewer, loved the content mate ! Curious to see you prompt inject a RAG model with knowledge stored as files (instead of SQL).

Hi, many thanks!

may i access the lab ftp ??? i want to practic that chall

The best explanation on PortSwigger XXE OOB out there.

Thanks for the video it is very helpful ! However, I still did not get why do you use a stack entity and why you do not directly use the exfil entity ?

Where can I get vulnerable code ?

thanks for sharing this man, I never could understand how this works without watching this

excellent quality I love so much you video man

I actually did this with a tuya iot camera and got a uboot shell by removing a 'silent' argument in the binary and flashing it on and i also set bootdelay to 3 in the binary and replaced the password hash with my own

Great work, thanks👏

great explanation!, you should make more portswigger videos🙌

great video!

You're one of my fav web app hacker. I cannot wait for the other videos.

*promosm* 😄

Which terminal to read uart?

Best resource, among all i have to check. Good Keep going

Thank you for taking the time to do this, to edit the videos and for sharing!

Another amazing video!

Finally your back 😁 and what a coincidence I was learning about JWTs only

Great Video! You're an awesome teacher! Keep pushing out videos!

Garr the literal wizard 🧙🔥🔥🔥

Long time no see, y'all! This video was recorded way back in 2022 and it covers the first Web Security Academy Lab. I added a bit of updates, but it should be a good "primer" for understanding JWTs and the "why" behind hacking them! We'll be covering the rest of the labs soon for a more in-depth, comprehensive methodology around JWTs, but we hope you enjoy this initial video!