It’s only been a few weeks since Google brought the Assistant to Google Maps to help you reply to messages, play music and more. This feature first launched in English and will soon start rolling out to all Assistant phone languages. In addition, Google also today announced that the Assistant will come to Android Messages, the standard text messaging app on Google’s mobile operating system, in the coming months.
If you remember Allo, Google’s last failed messaging app, then a lot of this will sound familiar. For Allo, after all, Assistant support was one of the marquee features. The different, though, is that for the time being, Google is mostly using the Assistant as an additional layer of smarts in Messages while in Allo, you could have full conversations with a special Assistant bot.
In Messages, the Assistant will automatically pop up suggestion chips when you are having conversations with somebody about movies, restaurants and the weather. That’s a pretty limited feature set for now, though Google tells us that it plans to expand it over time.
What’s important here is that the suggestions are generated on your phone (and that may be why the machine learning model is limited, too, since it has to run locally). Google is clearly aware that people don’t want the company to get any information about their private text chats. Once you tap on one of the Assistant suggestions, though, Google obviously knows that you were talking about a specific topic, even though the content of the conversation itself is never sent to Google’s servers. The person you are chatting with will only see the additional information when you push it to them.
Android malware developers are stepping up their billing fraud game with apps that disable Wi-Fi connections, surreptitiously subscribe users to pricey wireless services, and intercept text messages, all in a bid to collect hefty fees from unsuspecting users, Microsoft said on Friday.
This threat class has been a fact of life on the Android platform for years, as exemplified by a family of malware known as Joker, which has infected millions of phones since 2016. Despite awareness of the problem, little attention has been paid to the techniques that such “toll fraud” malware uses. Enter Microsoft, which has published a technical deep dive on the issue.
The billing mechanism abused in this type of fraud is WAP, short for wireless application protocol, which provides a means of accessing information over a mobile network. Mobile phone users can subscribe to such services by visiting a service provider’s web page while their devices are connected to cellular service, then clicking a button. In some cases, the carrier will respond by texting a one-time password (OTP) to the phone and requiring the user to send it back in order to verify the subscription request. The process looks like this:
The goal of the malicious apps is to subscribe infected phones to these WAP services automatically, without the notice or consent of the owner. Microsoft said that malicious Android apps its researchers have analyzed achieve this goal by following these steps:
Disable the Wi-Fi connection or wait for the user to switch to a mobile network
Silently navigate to the subscription page
Auto-click the subscription button
Intercept the OTP (if applicable)
Send the OTP to the service provider (if applicable)
Cancel the SMS notifications (if applicable)
Malware developers have various ways to force a phone to use a cellular connection even when it’s connected to Wi-Fi. On devices running Android 9 or earlier, the developers can invoke the setWifiEnabled method of the WifiManager class. For versions 10 and above, developers can use the requestNetwork function of the ConnectivityManager class. Eventually, phones will load data exclusively over the cellular network, as demonstrated in this image:
Once a phone uses the cellular network for data transmission, the malicious app surreptitiously opens a browser in the background, navigates to the WAP subscription page, and clicks a subscribe button. Confirming the subscription can be tricky because confirmation prompts can come by SMS, HTTP, or USSD protocols. Microsoft lays out specific methods that malware developers can use to bypass each type of confirmation. The Microsoft post then goes on to explain how the malware suppresses periodic messages that the subscription service may send the user to remind them of their subscription.
“By subscribing users to premium services, this malware can lead to victims receiving significant mobile bill charges,” Microsoft researchers wrote. “Affected devices also have increased risk because this threat manages to evade detection and can achieve a high number of installations before a single variant gets removed.”
Google actively bars apps from its Play market when it detects signs of fraud or malice, or when it receives reports of malicious apps from third parties. While Google often doesn’t remove malicious apps until after they have infected millions of users, apps downloaded from Play are generally regarded as more trustworthy than apps from third-party markets.
Researchers have identified stealthy new malware that threat actors have been using for the past 15 months to backdoor Microsoft Exchange servers after they have been hacked.
Dubbed SessionManager, the malicious software poses as a legitimate module for Internet Information Services (IIS), the web server installed by default on Exchange servers. Organizations often deploy IIS modules to streamline specific processes on their web infrastructure. Researchers from security firm Kaspersky have identified 34 servers belonging to 24 organizations that have been infected with SessionManager since March 2021. As of earlier this month, Kaspersky said, 20 organizations remained infected.
Stealth, persistence, power
Malicious IIS modules offer an ideal means to deploy powerful, persistent, and stealthy backdoors. Once installed, they will respond to specifically crafted HTTP requests sent by the operator instructing the server to collect emails, add further malicious access, or use the compromised servers for clandestine purposes. To the untrained eye, the HTTP requests look unremarkable, even though they give the operator complete control over the machine.
“Such malicious modules usually expect seemingly legitimate but specifically crafted HTTP requests from their operators, trigger actions based on the operators’ hidden instructions if any, then transparently pass the request to the server for it to be processed just like any other request,” Kaspersky researcher Pierre Delcher wrote. “As a result, such modules are not easily spotted by usual monitoring practices: they do not necessarily initiate suspicious communications to external servers, receive commands through HTTP requests to a server that is specifically exposed to such processes, and their files are often placed in overlooked locations that contain a lot of other legitimate files.”
Once SessionManager is deployed, operators use it to profile the infected environment further, gather passwords stored in memory, and install additional tools, including a PowerSploit-based reflective loader, Mimikat SSP, ProcDump, and a legitimate Avast memory dump tool. Kaspersky obtained multiple SessionManager variants that date back to at least March 2021. The samples show a steady evolution that has added more features with each new version. The most recent version of the malicious module includes the following:
Command name (SM_SESSION cookie value)
Command parameters (additional cookies)
FILEPATH: path of file to be read. FILEPOS1: offset at which to start reading, from file start.
FILEPOS2: maximum number of bytes to read.
Read the content of a file on the compromised server and send it to the operator as an HTTP binary file named cool.rar.
FILEPATH: path of file to be written.
FILEPOS1: offset at which to start writing.
FILEPOS2: offset reference.
FILEMODE: requested file access type.
Write arbitrary content to a file on the compromised server. The data to be written in the specified file is passed within the HTTP request body.
FILEPATH: path of file to be deleted.
Delete a file on the compromised server.
FILEPATH: path of file to be measured.
Get the size (in bytes) of the specified file.
Run an arbitrary process on the compromised server. The process to run and its arguments are specified in the HTTP request body using the format: <executable path>t<arguments>. The standard output and error data from process execution are sent back as plain text to the operator in the HTTP response body.
Check for SessionManager deployment. The “Wokring OK” (sic.) message will be sent to the operator in the HTTP response body.
S5HOST: hostname to connect to (exclusive with S5IP).
S5PORT: offset at which to start writing.
S5IP: IP address to connect to if no hostname is given (exclusive with S5HOST).
S5TIMEOUT: maximum delay in seconds to allow for connection.
Connect from compromised host to a specified network endpoint, using a created TCP socket. The integer identifier of the created and connected socket will be returned as the value of the S5ID cookie variable in the HTTP response, and the status of the connection will be reported in the HTTP response body.
S5ID: identifier of the socket to write to, as returned by S5CONNECT.
Write data to the specified connected socket. The data to be written in the specified socket is passed within the HTTP request body.
S5ID: identifier of the socket to read from, as returned by S5CONNECT.
Read data from the specified connected socket. The read data is sent back within the HTTP response body.
S5ID: identifier of the socket to close, as returned by S5CONNECT.
Terminate an existing socket connection. The status of the operation is returned as a message within the HTTP response body.
SessionManager gets installed after threat actors have exploited vulnerabilities known as ProxyLogon within Microsoft Exchange servers. Kaspersky has found it infecting NGOs, governments, militaries, and industrial organizations in Africa, South America, Asia, and Europe.
Kaspersky said it has medium-to-high confidence that a previously identified threat actor that researchers call Gelsemium has been deploying SessionManager. Security firm ESET published a deep dive on the group (PDF) last year. Kaspersky’s attribution is based on the overlap of code used by the two groups and victims targeted.
Disinfecting servers that have been hit by SessionManager or similar malicious IIS modules is a complicated process. Kaspersky’s post contains indicators that organizations can use to determine if they’ve been infected and steps they should take in the event they’ve been infected.
Chinese university students have been lured to work at a secretive technology company that masked the true nature of their jobs: researching western targets for spying and translating hacked documents as part of Beijing’s industrial-scale intelligence regime.
The Financial Times has identified and contacted 140 potential translators, mostly recent graduates who have studied English at public universities in Hainan, Sichuan and Xi’an. They had responded to job adverts at Hainan Xiandun, a company that was located in the tropical southern island of Hainan.
The application process included translation tests on sensitive documents obtained from US government agencies and instructions to research individuals at Johns Hopkins University, a key intelligence target.
Hainan Xiandun is alleged by a 2021 US federal indictment to have been a cover for the Chinese hacking group APT40. Western intelligence agencies have accused APT40 of infiltrating government agencies, companies and universities across the US, Canada, Europe and the Middle East, under the orders of China’s Ministry of State Security (MSS).
The FBI sought to disrupt the activities of Hainan Xiandun last July by indicting three state security officials in Hainan province—Ding Xiaoyang, Cheng Qingmin and Zhu Yunmin—for their alleged role in establishing the company as a front for state-backed espionage. Another man mentioned in the indictment, Wu Shurong, is believed to be a hacker who helped supervise employees at Hainan Xiandun.
Western intelligence services also seek out prospective spies from universities, with applicants undergoing rigorous vetting and training before joining the likes of the CIA in the US or the UK’s GCHQ signals intelligence agency.
But Chinese graduates targeted by Hainan Xiandun appear to have been unwittingly drawn into a life of espionage. Job adverts from the company were posted on university websites for translators without further explanation of the nature of the work.
This could have life-long consequences, as individuals identified as having co-operated with the MSS through their work for Hainan Xiandun are likely to face difficulty in living and working in western countries, a key motivation for many students who study foreign languages.
The FT contacted all 140 individuals on a leaked list of candidates compiled by security officials in the region to corroborate the authenticity of the applications. Several of those contacted initially confirmed their identities, but ended phone calls after being asked about their links to Hainan Xiandun. A few discussed their experience of the hiring process.
Their applications provide insight into the tactics of APT40, known for targeting biomedical, robotics and maritime research institutions as part of wider efforts to gain knowledge of western industrial strategy and steal sensitive data.
Hacking on that scale requires a huge workforce of English speakers who can help identify hacking targets, cyber technicians who can access adversaries’ systems and intelligence officers to analyze the stolen material.
Zhang, an English language graduate who applied to Hainan Xiandun, told the FT that a recruiter had asked him to go beyond conventional translation duties by researching the Johns Hopkins Applied Physics Laboratory, with instructions to find out information on the institution, including the CVs of the directors on its board, the building’s architecture and details of research contracts it had struck with clients.
The APL, a big recipient of US Department of Defense research funds, is likely to be of significant intelligence interest to Beijing and the individuals who work there prime hacking targets.
The instruction document asked the job candidates to download “software to get behind the Great Firewall.” It warns that the research will involve consulting websites such as Facebook, which is banned in China and so requires a VPN, software that masks the location of the user in order to gain access.
“It was very clear that this was not a translation company,” said Zhang, who decided against continuing with his application.
Dakota Cary, an expert in Chinese cyber espionage and former security analyst at Georgetown University, said the student translators were likely to be helping with researching organizations or individuals who might prove to be fruitful sources of sensitive information.
“The fact that you’re going to have to use a VPN, that you will need to be doing your own research and you need good language skills, all says to me that these students will be identifying hacking targets,” he said.
Cary, who testified earlier this year to the US-China economic and security review commission on Beijing’s cyber capabilities, said the instruction to investigate Johns Hopkins was an indicator of the level of initiative and ability to acquire specialist knowledge that the translators were expected to demonstrate.
One security official in the region said the revelations were evidence that the MSS was using university students as a “recruitment pipeline” for its spying activities.
Antony Blinken, US secretary of state, has previously condemned the MSS for building an “ecosystem of criminal contract hackers” who engage in both state-sponsored activities and financially motivated cyber crime. Blinken added that these hackers cost governments and businesses “billions of dollars” in stolen intellectual property, ransom payments and cyber defenses.
Hainan Xiandun asked the applicants to translate a document from the US Office of Infrastructure Research and Development containing technical explanations on preventing corrosion on transport networks and infrastructure. This appeared to test prospective employees’ abilities to interpret complex scientific concepts and terminology.
“It was a very weird process,” said Cindy, an English language student from a respected Chinese university. “I applied online and then the HR person sent me a highly technical test translation.” She decided against continuing with the application.
Adam Kozy, a former FBI official who worked most recently at cyber security company CrowdStrike, said he had not heard of western intelligence enlisting university students without them being given security clearance to collect intelligence.
“The MSS do everything very informally and they like the gray areas,” he said. “It’s interesting to see that they’re relying on a young student workforce to do a lot of the dirty work that may have those knock-on consequences later in life and most likely are not fully explaining those potential risks.”
The MSS did not respond to requests for comment.
Hainan Xiandun solicited applications on university recruitment sites and appears to have a close relationship with Hainan University. The company was registered on the first floor of the university library, home to the student computer room.
One job advert posted on the university’s foreign languages department website called for applications from English-speaking female students and Communist party members. The advert has been deleted since the FT’s queries regarding this story.
Several student applicants to Hainan Xiandun had won school prizes for their language skills and others held the added distinction of holding party membership.
According to the FBI’s indictment, MSS officers “co-ordinated with staff and professors at universities in Hainan and elsewhere in China” to further their intelligence goals. Personnel at one Hainan-based university also helped support and manage Hainan Xiandun as a front company, “including through payroll, benefits and a mailing address,” the indictment reads.
While the FBI accused the university of assisting the MSS in identifying and recruiting hackers and linguists to “penetrate and steal” from computer networks, it does not mention the university’s role in commandeering students to help the cause.
In response to the FT’s findings, Michael Misumi, chief information officer at Johns Hopkins APL, said that “like many technical organizations” the APL “must respond to many cyber threats and takes appropriate measures to continuously defend itself and its systems.”
Hainan University did not respond to requests for comment.
Applicants’ names have been changed to protect their identities