Get the highlights in your inbox every week.
Facebook open sources algorithms to find harmful content, Apple joins the Data Transfer Project, and more open source news
The fastest open source CPU ever, Facebook shares AI algorithms fighting harmful content, and more news
Catch up on the biggest open source headlines from the past two weeks.
In this edition of our open source news roundup, we share Facebook's choice to open source two algorithms for finding harmful content, Apple's new role in the Data Transfer Project, and more news you should know.
Facebook open sources algorithms used to find harmful content
Facebook announced that it has open sourced two algorithms used to find child exploitation, threats of terrorism, and graphic violence on the platform. In a blog post on August 1st, Facebook shared that PDQ and TMK+PDQF–two technologies that store files as digital hashes, then compare them with known examples of harmful content–are now live on GitHub.
The code release comes amidst increased pressure to get harmful content off the platform as soon as possible. After March's mass murder in New Zealand was streamed on Facebook Live, the Australian government threatened Facebook execs with fines and jail time if the video wasn't promptly removed. By releasing the source code for these algorithms, Facebook said that it hopes nonprofits, tech companies, and solo developers can all help them find and remove harmful content faster.
Alibaba launches the fastest open source CPU
Pingtouge Semiconductor - an Alibaba subsidiary - announced its Xuantie 91 processor last month. It's equipped to manage infrastructure for AI, the IoT, 5G, and autonomous vehicles, among other projects. It boasts a a 7.1 Coremark/MHz, making it the fastest open source CPU on the market.
Pintogue announced plans to make its polished code available on GitHub this September. Analysts view this release as a power move to help China hit its goal of using local suppliers to meet 40 percent of processor demand by 2021. Recent tariffs on behalf of the U.S. threaten to derail this goal, creating the need for open source computer components.
Mattermost makes the case for open source collaboration apps
All open source communities benefit from one or more places to communicate with each other. The world of team chat apps seems dominated by minimal, mighty tools like Slack and Microsoft Teams. Most options are cloud-based and proprietary; Mattermost takes a different approach by selling the value of open source collaboration apps.
“People want an open-source alternative because they need the trust, the flexibility and the innovation that only open source is able to deliver,” said Ian Tien, co-founder and CEO of Mattermost.
With clients that range from Uber to the Department of Defense, Mattermost cornered a crucial market: Teams that want open source software they can trust and install on their own servers. For businesses that need collaboration apps to run on their internal infrastructure, Mattermost fills a gap that Atlassian left bare. In an article for Computerworld, Matthew Finnegan explores why on-premises, open source chat tools aren't dead yet.
Apple joins the open source Data Transfer Project
Google, Facebook, Twitter, and Microsoft united last year to create the Data Transfer Project (DTP). Hailed as a way to boost data security and user agency over their own data, the DTP is a rare show of solidarity in tech. This week, Apple announced that they'll join the fold.
The DTP's goal is to help users transfer their data from one online service to another via an open source platform. DTP aims to take out the middleman by using APIs and authorization tools to let users transfer their data from one service to another. This would erase the need for users to download their data, then upload it to another service. Apple's choice to join the DTP will allow users to transfer data in and out of iCloud, and could be a big win for privacy advocates.
In other news
Thanks, as always, to Opensource.com staff members and moderators for their help this week.