Fidaro is live.
It's an AI you can actually talk to in private. Your messages are encrypted on your device before they leave your browser. We pass them along but can't read them.
No logging. No storing. No profile.
Free in beta, try it now: https://t.co/svWgSuPCc1
New York State is warning people not to share passwords, cc numbers, medical records, or legal documents with AI chatbots.
Why? Because many AI services can store and access your conversations.
Good advice. But concerning that this warning is even necessary.
Your AI probably knows things even your closest friends don't.
Your 3am worries. Your biggest secrets. The questions you're afraid to ask anyone else.
At some point, your AI chats become a personal diary.
And diaries deserve a lock. 🔒
Think of the last AI chat you shared. You sent it to one person. Someone you trust.
You probably never thought about the title on that link.
@Meta Muse users didn't either. Researchers found 6,800+ of their shared chats just by searching DNS records for those titles.
https://t.co/2mPYZbVp8K
Would you be comfortable sharing your entire AI chat history with your coworkers? Your employer? Your family?
Probably not.
So why are we so comfortable letting companies store all of it?
144,000+ AI conversations left exposed online. Some contained people's home addresses.
They weren't exposed by @ChatGPT, @Claude or @Gemini, but by https://t.co/35LYsPnRtt, an app that offered access to all three.
You chose the AI. But who else is keeping a copy of everything you tell it?
https://t.co/7YjbMrK7d7
Even @ChatGPT's Temporary Chats may be stored for up to 30 days.
That's 30 days where conversations you thought were temporary could still exist on company servers.
Fidaro works differently.
Your messages are encrypted before they leave your device and decrypted only inside a secure, isolated environment. We can't read your conversations, and your chat history is stored encrypted.
Same questions. Same AI experience. But your private conversations stay private.
Imagine finding out your child's favorite toy has been saving everything they've told it.
A researcher discovered 13 MILLION conversations stored by an AI robot for kids, despite promises to delete voice recordings.
Children's names, birthdays, even emotional vulnerabilities were exposed.
And researchers could remotely manipulate what the robots said.
Would you still let your child talk to one?
Before you tell an AI something personal, ask 3 questions:
🔒 Who can read it?
💾 What's stored?
👀 Who else gets it?
Fidaro: chats are encrypted on your device and only decrypted in a secure, isolated environment. Your history stays encrypted. No readable logs. Nothing sent to the creators of the models.
Not even we can read your chats.
“Glamorous wide fit”
“ASOS petite”
Imagine getting a message mentioning something you searched for on ASOS.
Hackers stole customer names, contact details and even search history.
They also sent notifications through the REAL ASOS app.
⚠️ Just because a message knows what you searched for doesn't mean it's legit.
https://t.co/4qDXVpVfX9
🚩 If a chatbot tells you to paste a command into PowerShell, stop.
Huntress found malicious Custom GPTs on @ChatGPT that sent users to a fake CAPTCHA.
The “fix” installed malware that could watch your screen, access your files, and turn on your camera and mic.
Source: https://t.co/ouiIJgwfdE
Your AI chat shouldn’t be feeding trackers.
IMDEA tested 9 major chatbots. 6 web clients sent conversation URLs, titles, prompts or screenshots to third parties.
Fidaro is built so even we can’t read your conversations.
Free during beta 👇
16,000+ app databases were found exposed online.
One valet app leaked 100,000 customers’ license plates and visit history. An immigration service exposed 884 passwords in plain text.
You can’t audit every app you use. The less personal data they need from you, the better.
https://t.co/2klRwl2UUX
When you upload a photo to an AI chat, you probably don't expect it to end up on a public image-hosting site.
@OpenAI says its agents leaked 53 user images from anonymized training data.
Fidaro doesn't use your chats for training. We can't read them in the first place.
"I have nothing to hide" assumes privacy is only for secrets.
Your salary isn't a secret. Neither are your health questions, relationship problems, or work issues.
You still probably wouldn't want to say them out loud on a crowded train.
Privacy isn't about hiding. It's about choosing who gets to hear the conversation.
Fidaro keeps your AI conversations out of every room you did not choose.
.@OpenAI pays hundreds of contractors over $50 an hour to read real @ChatGPT chats and rate the answers.
No usernames. But they can see the full chat and a summary of your memory.
Training is on by default for Free, Plus and Pro. Turning it off does not remove past chats from training.