Why is no one talking about this? This is why I don't use an AI browser You can literally get prompt injected and your bank account drained by doomscrolling on reddit:
Why is no one talking about this? This is why I don't use an AI browser You can literally get prompt injected and your bank account drained by doomscrolling on reddit: https://t.co/aGN8xrdZtD
No one seems to be concerned about this, it seems to me like the #1 problem with any agentic AI stuff You can get pwned so easily, all an attacker has to do is literally write words down somewhere???
@zack_overflow nobody is talking about this because nobody uses fucking AI browsers
@zack_overflow This is why I don't have a bank account 👀
@zack_overflow it does seem like basically every AI product these days is made by people who are fully technologically illiterate and/or have no qualms about putting their users in danger
@zack_overflow You’re looking at this all wrong Zach. Now you need to fundraise for an AI browser shield plugin that prevents prompt injections 🫡😂
@zack_overflow Social engineering has always been the easiest way to "hack" someone. Except now, that "someone" is a completely brainless AI agent. Oops!
@zack_overflow They are you're just not subscribed. youtube.com/watch?v=nmkk_b…
@zack_overflow I got 99 problems, letting my agentic browser auth as my main is not one.
@zack_overflow Why are you looking at your bank account on your laptop? Use the app like an adult
@zack_overflow I want AI available in my browser. When *I* need it. But I don't want "always on" AI browser. I actually don't want always on AI anything.
@zack_overflow This is how agentic browsers will improve gross margins
@zack_overflow You can add fun and gasoline to the fire by reviewing their privacy policies. perplexity.ai/hub/legal/come… perplexity.ai/hub/legal/priv…
@zack_overflow Uncontained AI is horrible for security. Businesses need to be extra careful while integrating AI tools. It's probably 10x easier to socially engineer an AI agent than it is a human.
@zack_overflow how easy can you rectify this? can you just tell ai via system prompt not to trust website content?
@zack_overflow that's why you close your bank account and buy a hardware wallet
Giving an agent full access to your data without clear guardrails is a really bad idea. We automate checkouts for e-commerce stores and work with very sensitive information, but our agents never see the real data. They only fill forms with placeholders, which later get swapped with the actual values downstream. Prompt injection is a real risk, and while the industry will adapt, you need to be extremely cautious when letting agents operate in these contexts. Long story short: do not give "admin" privileges to AI Agents in the wild.
@zack_overflow fell for it again award for the vibetards
@zack_overflow Crypto guard from @WalrusProtocol fixes this or nah @0xd34th ? ( Asking fr )
@zack_overflow Yeah…also I’ve found them pretty awkward to use anyway.
@zack_overflow A few years ago we were talking about AI safety all the time. OpenAI leadership fell apart due to this. Fast forward to today we're giving AI control over everything. Strange
@zack_overflow Cause nobody is using Comet or any other ai chrome reskins