Post

@shoshanazuboff

More from @ProtonPrivacy

Proton

@ProtonPrivacy

Apr 10

🧵 241 official German parliament member email addresses were found on the dark web -- that's 13% out of the 1,874 checked. Leaked via breaches of services like Dropbox, LinkedIn & even adult sites.
1/5

⬇️

We found that members of state parliament had more than just their email addresses leaked. There were also birthdates, addresses, passwords (153 in plaintext), and more, all for sale on the dark web.
2/5

⬇️

One MP, in Nordrhein-Westfalen, had their gov email included in 24 breaches. One in Rheinland-Pfalz had 14 plaintext passwords exposed - the most of any German politician.
3/5

⬇️

Read 5 tweets

Proton

@ProtonPrivacy

Apr 9

Today, we're sharing plans for the next 3-6 months — new features & improvements to give you more powerful tools to protect privacy, boost productivity, & take control.

We work for you and only you, so we selected these features based on your feedback & requests. 🧵

1/8

@ProtonVPN

Popular features will be available on more platforms.

Exclude apps & websites from your VPN with Split Tunneling on Mac & Linux.

Get the parental controls & filtering from 3rd-party DNS with Custom DNS on iPhone & Mac.

Explore more:

2/8protonvpn.com/blog/vpn-roadm…

@ProtonMail

When replying to emails, finding the right message often takes the most time.

Category View filters out the noise so you can focus on the important emails.

Stop endless scrolling. Quickly surface the emails you need.

Explore more:

3/8proton.me/blog/mail-cale…

Read 8 tweets

Proton

@ProtonPrivacy

Apr 1

Meta AI in WhatsApp is now near-inescapable, with the feature rolling out to more devices.

Even users based in one of the 41 European countries should soon see it if they are using the messaging app.

This AI is a less-powerful version of Meta AI's web app.

1/5

Users should see an AI button above the one used to create a new chat, and prompt shortcuts in the search bar.

This AI can be used to create stickers, but not to transcribe audio or summarize the contents of group chats. It can also share Bing links and generate text.

2/5

Crucially, it is not possible to turn off Meta AI in WhatsApp.

Meta states that chats will not be used to train the AI models, but Meta AI conversations could be used to train future versions.

If you don't trust this statement from Meta, there are steps you can take.

3/5

Read 5 tweets

Proton

@ProtonPrivacy

Mar 28

How we use the internet is largely shaped by a handful of US-based tech giants.

US gov't surveillance laws allow them to demand access to your data without your knowledge or a warrant.

Read the thread below for the list of European alternatives that put your privacy 1st.

1/16

We founded Proton in 2014 specifically to resist mass surveillance and abuses by Big Tech.

Google’s Gmail, Drive, Docs, Password Manager and Calendar can now be replaced with @ProtonMail, @ProtonDrive, Proton Docs, @Proton_Pass and Proton Calendar.

2/16

🔍 When it comes to search engines there are:

🇫🇷 @Qwant_FR, a privacy-first search engine based on Bing;
🇩🇪 @ecosia, a search engine that funds planting trees with ad revenue; and
🇬🇧 @mojeek, an independent search engine with strict privacy policies.

3/16

Read 17 tweets

Proton

@ProtonPrivacy

Mar 27

Think this is a fun trend? Think again.

While some don't have an issue sharing selfies on social media, the trend of creating a "Ghibli-style" image has seen many people feeding OpenAI photos of themselves and their families.

Here's why that's a problem:

1/4

Aside from the risks of data breaches, once you share personal photos with AI, you lose control over how they are used, since those photos are then used to train AI.

For instance, they could be used to generate content that may be defaming or used as harassment.

2/4

Many AI models, particularly those used in image generation, rely on large training datasets. In some cases, photos of you, or with your likeness, might be used without your consent.

Lastly, your data could be used for personalized ads and/or sold to third parties.

3/4

Read 4 tweets

Proton

@ProtonPrivacy

Mar 26

None of your business (noyb), a privacy non-profit based in Austria and Proton Lifetime Fundraiser beneficiary, is helping a Norwegian ChatGPT user in a complaint against OpenAI.

This complaint has come as a result of fake information being generated by their LLM.

1/6

The user tried to find out if OpenAI had information about him, and so asked it to tell him who he was, using his full name.

ChatGPT confidently made up a story where he was a convicted child murderer, citing the right city and correctly stating that he has three kids.

2/6

Noyb undertook research in order to try and understand why this fabricated information had made it into the output, but there was no explanation or similar stories which could've been mixed in.

LLMs work by predicting the next word in a sequence in response to user input.

3/6

Read 6 tweets

Share this page!

Enter URL or ID to Unroll

Proton

Try unrolling a thread yourself!

More from @ProtonPrivacy

Proton

Proton

Proton

Proton

Proton

Proton

Did Thread Reader help you today?

Don't want to be a Premium member but still want to support us?

Send Email!