We’ve written about the Swiss company Proton’s moves to take on Google and Microsoft with an expanding variety of privacy-focused internet services, and the company is announcing yet another new tool today. Proton Meet, as the name suggests, is a video-calling service that sounds comparable to Zoom, Microsoft Teams and, naturally, Google Meet.

As with everything Proton does, Meet is end-to-end encrypted, but the company is taking extra steps towards security and anonymity here. You don’t need a Proton account to join a Meet call, which should help service gain traction — you can use Proton and not totally throw off everyone else who’s still using other systems. Proton says that Meet can be used anonymously and no logs are kept. The company even says that a Proton account isn’t needed to host a meeting. If you visit the Proton Meet site, you can start an anonymous call with up to four participants for free.

Proton says that anyone with an account (even a free one) can start Meet call with up to 50 participants for up to one hour, but it is also offering a dedicated Meet Professional plan for $8 a month with fewer restrictions (though we don’t have the specifics yet).

Meet joins the other core Proton tools including Mail, Calendar, VPN, Drive and a password manager. Drive in particular has gotten some notable upgrades recently — it offers collaborative documents and spreadsheets, just like Google. With the addition of Meet, Proton has most of the core services that Google offers covered. There’s even a little AI bot if that’s your thing.

Proton is using this launch as an opportunity to rebrand its services, gathering them all under the new Proton Workspace (another name pulled directly from Google’s offerings). In addition to the dedicated Meet Professional plan, Proton is offering Workplace Standard for $13 per month (billed annually) or $15 a month (billed monthly). That includes the aforementioned tools and 500GB of space by default. There’s also a Premium plan ($20/month annually or $25/month billed monthly) that adds more storage, access to the Lumo chatbot, expanded Meet participant limits and a few other tidbits.

The company says it’ll continue offering its other plans, including a Mail-only option that costs $36 a year as well as other bundles for an individual, two users or a whole family. But Proton is obviously looking to expand its business client base with Meet in addition to growing its user base (currently more than 100 million). Proton says it has about 100,000 enterprise customers, and adding a tool like Meet to its portfolio should make it easier to court companies as well as individuals.

This article originally appeared on Engadget at https://www.engadget.com/apps/proton-adds-a-secure-video-conferencing-service-called-meet-to-its-toolbox-121729143.html?src=rss

©

Match Group and its subsidiary OkCupid has finally settled a lawsuit with the Federal Trade Commission that dates back to its alleged sharing of user data back in 2014. According to the lawsuit, the FTC accused OkCupid of inappropriately sharing personal user data that includes photos and location info with a third party company, Clarifai, which offers AI-powered software for uses like facial recognition and content moderation.

According to the FTC, OkCupid's privacy policy at the time noted that the company wouldn't share a user's personal information with others, except for some cases including "service providers, business partners, other entities within its family of businesses." However, the lawsuit accused OkCupid of sharing three million photos of its users to Clarifai, which the FTC claims is a "unrelated third party" that didn't fall under the allowed entities. On top of that, the lawsuit alleged that OkCupid didn't inform its users of this data sharing, nor give them a chance to opt out.

"While we do not admit any wrongdoing, we have settled this matter with the FTC with no monetary penalty to resolve an issue from 2014 and move forward," an OkCupid spokesperson told Engadget, adding that the allegations don't reflect how OkCupid operates today. "Over the years, we have further strengthened our privacy practices and data governance to ensure we meet the expectations of our users."

Moving forward, the settlement would "permanently prohibit" Match Group, which owns OkCupid, and Humor Rainbow, which operates OkCupid, from misrepresenting what kind of personal information it collects, the purpose for collecting the data and any consumer choices to prevent data collection. Even after the 2014 incident, OkCupid was found with security flaws that could've exposed user account info but, which were quickly patched in 2020.

This article originally appeared on Engadget at https://www.engadget.com/cybersecurity/okcupid-settles-ftc-case-on-alleged-misuse-of-its-users-personal-data-175159228.html?src=rss

©

© OkCupid

Artificial intelligence companies have been working at breakneck speeds to develop the best and most powerful tools, but that rapid development hasn't always been coupled with clear understandings of AI's limitations or weaknesses. Today, Anthropic released a report on how attackers can influence the development of a large language model.

The study centered on a type of attack called poisoning, where an LLM is pretrained on malicious content intended to make it learn dangerous or unwanted behaviors. The key finding from this study is that a bad actor doesn't need to control a percentage of the pretraining materials to get the LLM to be poisoned. Instead, the researchers found that a small and fairly constant number of malicious documents can poison an LLM, regardless of the size of the model or its training materials. The study was able to successfully backdoor LLMs based on using only 250 malicious documents in the pretraining data set, a much smaller number than expected for models ranging from 600 million to 13 billion parameters. 

"We’re sharing these findings to show that data-poisoning attacks might be more practical than believed, and to encourage further research on data poisoning and potential defenses against it," the company said. Anthropic collaborated with the UK AI Security Institute and the Alan Turing Institute on the research.

This article originally appeared on Engadget at https://www.engadget.com/researchers-find-just-250-malicious-documents-can-leave-llms-vulnerable-to-backdoors-191112960.html?src=rss

©

Last month, Alphabet told the House Judiciary Committee that it would allow some creators who were banned from YouTube for spreading COVID-19 and election-related misinformation to come back to the platform. Now, the company is outlining how that process will work, though it's not clear exactly who will be provided that opportunity,

YouTube says that beginning today it will give "some previously terminated creators" the ability to "request" a fresh channel in an effort the company is characterizing as a bid for "second chances" on the platform. The update doesn't reference its COVID-19 or election misinformation policies, which had come under fire from House Republicans. Instead, YouTube says that the "pilot program" will be available to some "qualified creators." It doesn't explain how it will determine who qualifies. Last month, far-right personalities Nick Fuentes and Alex Jones — both of whom were banned from YouTube years before the pandemic — attempted to rejoin YouTube after the letter was made public and were promptly removed.

"Our goal is to roll this out to creators who are eligible to apply over the coming months, and we appreciate the patience as we ramp up, carefully review requests, and learn as we go," the company wrote. "Not every type of channel termination will be eligible." Among the factors YouTube will consider are "whether the creator committed particularly severe or persistent violations of our Community Guidelines or Terms of Service, or whether the creator’s on- or off-platform activity harmed or may continue to harm the YouTube community." People banned for copyright infringement won't be able to apply. 

All that is still remarkably vague compared to the language Alphabet used in its letter to the House Judiciary Committee last month. "YouTube will provide an opportunity for all creators to rejoin the platform if the Company terminated their channels for repeated violations of COVID-19 and elections integrity policies that are no longer in effect," it said. 

Importantly, YouTube isn't giving previously banned creators access to their old channels and subscribers. Anyone who is let back on under the program will instead have the opportunity to create an entirely new channel, which means they'll have to rebuild their presence from scratch.

This article originally appeared on Engadget at https://www.engadget.com/big-tech/youtube-will-give-second-chances-to-some-banned-creators-172000443.html?src=rss

©

California has passed three new bills designed to boost privacy for internet users, governor Gavin Newsom's office announced. The biggest one, AB 566, builds on a 2018 law by requiring web browsers to let users universally opt-out of allowing third parties to sell their data.

The original California Consumer Privacy Act from 2018 only let Californians opt out third-party data sharing one site at a time. However, AB 566 signed into law yesterday by Newsom requires web browsers like Chrome, Firefox and Safari to allow users to opt out of all third-party tracking with a single setting. "This law will help people protect their personal data by allowing them to simply switch a toggle that tells businesses they can’t sell or share it," said Consumer Reports policy analyst Matt Schwartz. 

The bill was originally passed by the California legislature last month, but its signing by the governor wasn't necessarily a done deal. Newsom vetoed a similar bill last year for being overly broad as it also applied to smartphone operating systems. He also said that major browsers already offer one-click opt out for third-party data sharing, though Consumer Watchdog said at the time that none offer a universal way to decline data sharing. 

Two other bills will also help internet users keep their data to themselves. SB 361 boosts the Data Broker Registration Law (Delete Act) signed into law in October 2023 by giving consumers more information about which personal information is collected by data brokers and who else might have it. AB 656, meanwhile, requires social media companies to make canceling an account straightforward and clear while it triggers full deletion of the user’s personal data.

This article originally appeared on Engadget at https://www.engadget.com/big-tech/california-just-passed-three-bills-to-boost-internet-privacy-120031025.html?src=rss

©

© NurPhoto via Getty Images