Perplexity (@perplexity_ai)
Comet가 기업용 AI 브라우저 제품 'Comet Enterprise'를 정식 출시했습니다. 엔터프라이즈 팀을 대상으로 브라우저 내에서 연구, 작업 자동화 및 업무 수행이 가능하도록 설계된 새로운 제품 발표입니다.
Perplexity (@perplexity_ai)
Comet가 기업용 AI 브라우저 제품 'Comet Enterprise'를 정식 출시했습니다. 엔터프라이즈 팀을 대상으로 브라우저 내에서 연구, 작업 자동화 및 업무 수행이 가능하도록 설계된 새로운 제품 발표입니다.
What could go wrong when injecting AI into the web browser?
"Comet can be steered to extract a stored credential from a user’s 1Password vault and transmit it to an external attacker, operating entirely inside the user’s authenticated browser context."
In essence: If you want to get #hacked with #promptinjection via malicious content from web sites your ai browser reads and interprets ... go for #aibrowser
It may nicely serve prompts from malicious sites to e.g. send them your credit card number, stored credentials etc.
In simple words: quite close to jumping from the next bridge and expecting to stand up and walk away unharmed 🤪
------------ %< snip %< -------------
The lethal trifecta for AI agents: private data, untrusted content, and external communication
The other day I was thinking, when will infostealers start collecting information from "AI" browsers.
Today noticed that StealC has posted an update, where they added a feature of collecting Sigma AI Browser data.
''Sigma AI Browser is an AI‑first agentic browser that combines an AI agent, deep research, and AI tools to help you navigate, create, and'' sloooop
Anyways, left to find some time and energy to look for some stealc logs and see if it syphons some more juicy data from "ai" browsers.
2026: Wir leben in einer Welt, in der Überwachung und Datensammlung nicht nur die Norm sind, sondern als Privileg angesehen werden. Der Datenschutz, wie wir ihn einst kannten, ist nicht einfach nur geschwächt, er existiert nicht mehr. Die Grenzen der Privatsphäre sind längst verwischt. Du wirst nicht mehr gefragt, ob du einem Dienst deine Daten geben möchtest. Stattdessen wird dir erzählt, dass du „mit einer Datenfreigabe zur globalen Gemeinschaft - für ein besseres, gerechteres, sichereres und nachhaltigeres Miteinander - beiträgst.“
#daten #datenschutz #privacy #ueberwachung #privatsphaere #datenteilen #privileg #ki #ai #aibrowser #iot
OpenAI says AI browsers may always be vulnerable to prompt injection attacks
https://blog.jim-nielsen.com/2025/dont-use-ai-browsers-they-say
Le début du billet de Jim Nielsen’s Blog
I don’t have access to the original paper (because I’d have to pay Gartner for it), but the reporting on Gartner’s research says this:
research VP Dennis Xu, senior director analyst Evgeny Mirolyubov, and VP analyst John Watts observe “Default AI browser settings prioritize user experience over security.”
C’mon, let’s call a spade a spade: they prioritize their maker’s business model over #security
#aibrowser
#cybersecurity