ClusterFuzz to be Open Source Now

Program's code is written in Python and Go, and distributed under the Apache 2.0 license
08 February 2019   234

Google has opened the source code for the ClusterFuzz platform, intended for fuzzing code testing using a server cluster. In addition to coordinating the execution of checks, ClusterFuzz also automates the execution of tasks such as sending a notification to developers, creating an application for a patch (issue), tracking a bug fix, and closing reports after a patch. The code is written in Python and Go, and distributed under the Apache 2.0 license. ClusterFuzz instances can run on Linux, macOS and Windows systems, as well as in various cloud environments.

Since 2011, ClusterFuzz has been used in the depths of Google to detect errors in the Chrome codebase and to ensure the operation of the OSS-Fuzz project, in the framework of which continuous fuzzing testing of open source software was organized. In total, ClusterFuzz has revealed more than 16 thousand errors in Chrome and more than 11 thousand errors in 160 open source projects participating in the OSS-Fuzz program. Due to the continuous process of checking the current code base, errors are usually caught within a few parts after the code is introduced and the changes causing them.

OpenAI to Create Fake News Creating Algorithm

On the basis of one or two phrases that set the theme, it is able to “write” a fairly plausible story
18 February 2019   146

The GPT-2 algorithm, created by OpenAI for working with language and texts, turned out to be a master in creating fake news. On the basis of one or two phrases that set the theme, it is able to “compose” a fairly plausible story. For example:

  • an article about scientists who have found a herd of unicorns in the Andes;
  • news about pop star Miley Cyrus caught on shoplifting;
  • artistic text about Legolas and Gimli attacking the orcs;
  • an essay on how waste recycling harms the economy, nature, and human health.

The developers did not publish the source code of the model entirely, fearing abuse by unscrupulous users. For fellow researchers, they posted on GitHub a simplified version of the algorithm and gave a link to the preprint of the scientific article. The overall results are published on the OpenAI blog.

GPT-2 is a general purpose algorithm. The developers taught it to answer questions, “understand” the logic of a text, a sentence, finish building phrases. In this case, the algorithm worked worse than the model of a specific purpose. Researchers suggest that the indicators can be improved by expanding the training datasets and choosing computers more efficiently.