How an AI trained to read scientific papers could predict future discoveries

In the new study, an AI learned to retrieve information from scientific literature via unsupervised learning. This has remarkable implications. So far, most of the existing automated NLP-based methods are supervised, requiring input from humans. Despite being an improvement compared to a purely manual approach, this is still a labour intensive job. However, in the new study, the researchers created a system that could accurately identify and extract information independently. It used sophisticated techniques based on statistical and geometrical properties of data to identify chemical names, concepts and structures. This was based on about 1.5m abstracts of scientific papers on material science.

New Bug Found in NSA’s Ghidra Tool

A medium severity bug reported on Saturday impacts Ghidra, a free, open-source software reverse-engineering tool released by the National Security Agency earlier this year. The vulnerability allows a remote attacker to compromise exposed systems, according to a NIST National Vulnerability Database description. No fix is currently available. Despite the warning, researchers are downplaying the impact of the bug. They maintain conditions needed to exploit the flaw, tracked as CVE-2019-16941, are rare. They also note, the NSA’s GitHub repository for Ghidra indicates a patch is currently in the works.

Google is making it easier to check if your passwords have been compromised in a data breach

Google has a password manager that syncs across Chrome and Android, and now the company is adding a “password checkup” feature that will analyze your logins to ensure they haven’t been part of a massive security breach — and there have been oh so many of those. Password checkup was already available as an extension, but now Google is building it right into Google account controls. And it’ll be prominently featured at passwords.google.com, which is the URL shortcut to Google’s password manager.

NSA launches new cyber defense directorate

The National Security Agency on Tuesday will launch an organization to prevent cyberattacks ­on sensitive government and ­defense-industry computers — with an eye also toward helping shield critical private-sector systems. For decades the agency had a cyberdefense organization, the Information Assurance Directorate (IAD), that focused on safeguarding the government’s classified and sensitive networks, as well as the private sector’s, when asked. What is new, NSA officials said, is that the agency is hitching together under one roof threat detection, cyberdefense and ­future-technologies personnel. They are calling it the Cybersecurity Directorate.

Exim suffers another ‘critical’ remote code execution flaw

Remember the critical remote code execution (RCE) vulnerability in the Exim email server, CVE-2019-15846, from mid-September? Barely two weeks later, and the software’s maintainers have issued an advisory for another potentially troublesome bug, identified as CVE-2019-16928, which has been given the same critical rating. Affecting all Exim versions between and including 4.92 to 4.92.2, this one’s described as: A heap-based buffer overflow in string_vformat (string.c). The currently known exploit uses an extraordinary long EHLO string to crash the Exim process that is receiving the message.

FDA issues warning on medical devices that are vulnerable to takeover from hackers

Medical devices that use third-party, decades-old software called IPnet are at risk, the FDA said. The regulator said it’s not sure how many or even which specific devices, such as insulin pumps or pacemakers, are vulnerable to getting hacked. Researchers have identified 11 vulnerabilities that may allow “anyone to remotely take control of the medical device.”

Experts found 20 Million tax records for Russian citizens exposed online

Security experts from Comparitech along with security researcher Bob Diachenko discovered 20 million tax records belonging to Russian citizens exposed online in clear text and without protection. The experts found an unprotected Elasticsearch cluster that was containing personally identifiable information on Russian citizens spanning from 2009 to 2016. “A database of more than 20 million Russian tax records was found on an unsecured server, accessible to anyone with a web browser.” reads the post published by Comparitech. “Comparitech partnered with security researcher Bob Diachenko to investigate the data exposure, which included sensitive personal and tax information. The database was taken offline after Diachenko notified the owner, who is based in Ukraine.”

Study: Americans believe social networks make the news worse

You’re not alone if you think social networks are degrading the quality of news. A Pew study indicates that 62 percent of American adults believe social media companies have “too much control” over news, and 55 percent believe these companies have created a “worse mix” of news through their feed algorithms and other editorial choices. Not surprisingly, the survey respondents pinned it on the quality of the news that surfaced in their social streams.

US hospitals turn away patients as ransomware strikes

Three US hospitals have been forced to temporarily close their doors to “all but the most critical new patients” following a ransomware outbreak. “A criminal is limiting our ability to use our computer systems in exchange for an as-yet unknown payment,” said DCH Health System. DCH operates the three affected hospitals in Alabama. One cyber-security expert said the groups using ransomware were becoming increasingly well organised. Computers at the DCH Regional Medical Center in Tuscaloosa, Fayette Medical Center and Northport Medical Center were infected with ransomware.

Zendesk discloses 2016 data breach

Customer support ticketing platform Zendesk disclosed today a security breach dating back to November 2016. In a message posted on its website, the company said that a hacker accessed the personal information of approximately 10,000 users that had registered Zendesk Support and Chat accounts. Zendesk said it discovered the breach last week, on September 24, nearly three years after it took place. The company said it learned of the incident “from a third-party.” 

Molecule.one uses machine learning to make synthesizing new drugs a snap

Say you’re a pharmaceutical company. You’ve figured out that a novel molecule could be effective in treating an illness — but that molecule only exists in a simulation. How do you actually make it, and enough of it, to test in the real world? Molecule.one is a computational chemistry platform that helps bring theoretical substances to life, and it is debuting its product onstage at Disrupt SF Startup Battlefield.