Data-Nuking Prompt Injection
Failed to add items
Sorry, we are unable to add the item because your shopping cart is already at capacity.
Add to basket failed.
Please try again later
Add to wishlist failed.
Please try again later
Remove from wishlist failed.
Please try again later
Adding to library failed
Please try again
Follow podcast failed
Unfollow podcast failed
Data-Nuking Prompt Injection
-
Narrated by:
-
By:
Drew and Dan explore how a single line of text in a public repo can ruin your day, what happens when AI coding agents go off the rails near your data, and the thorny questions of AI safety and ethics. Google has bad news for back-button hijackers. Drew shares lessons learned and challenges from making his apps accessible, and the two get into how optimizing for the wrong metric can quietly steer AI somewhere nobody intended.
Chapters
- 00:00 Google Punishing Back-Button Hijackers
- 02:31 Data Nuking Prompt Injection
- 12:43 App Accessibility Lessons and Challenges
- 20:48 AI Metrics and Unintended Consequences
Finding Drew:
- Bluesky
- X / Twitter
- Drew’s YouTube live stream coding channel, @DrewBensonCodes
Finding Dan:
- Bluesky
- Dan’s Certainly Swift blog
Mentioned in this episode:
- Google Punishing Back-Button Hijacking - Google’s policy
- Data Nuking Prompt Injection
- Accessibility Nutrition Labels
- Drew’s MacOS Agent Smith agent harness
- Drew’s Sight Words: First Words Match app
- Drew’s FunVoice Reverse Audio Singing app
- Goodhart’s Law
Drew and Dan in the Morning can be found on:
- Apple Podcasts
- Spotify
- YouTube
- Overcast
- Deezer
adbl_web_anon_alc_button_suppression_t1
No reviews yet