Login
V'ger is an interactive command-line application for post-exploitation of authenticated Jupyter instances with a focus on AI/ML security operations.
FreshRSS pip install vgervger --helpCurrently, vger interactive has maximum functionality, maintaining state for discovered artifacts and recurring jobs. However, most functionality is also available by-name in non-interactive format with vger <module>. List available modules with vger --help.
Once a connection is established, users drop into a nested set of menus.
The top level menu is: - Reset: Configure a different host. - Enumerate: Utilities to learn more about the host. - Exploit: Utilities to perform direct action and manipulation of the host and artifacts. - Persist: Utilities to establish persistence mechanisms. - Export: Save output to a text file. - Quit: No one likes quitters.
These menus contain the following functionality: - List modules: Identify imported modules in target notebooks to determine what libraries are available for injected code. - Inject: Execute code in the context of the selected notebook. Code can be provided in a text editor or by specifying a local .py file. Either input is processed as a string and executed in runtime of the notebook. - Backdoor: Launch a new JupyterLab instance open to 0.0.0.0, with allow-root on a user-specified port with a user-specified password. - Check History: See ipython commands recently run in the target notebook. - Run shell command: Spawn a terminal, run the command, return the output, and delete the terminal. - List dir or get file: List directories relative to the Jupyter directory. If you don't know, start with /. - Upload file: Upload file from localhost to the target. Specify paths in the same format as List dir (relative to the Jupyter directory). Provide a full path including filename and extension. - Delete file: Delete a file. Specify paths in the same format as List dir (relative to the Jupyter directory). - Find models: Find models based on common file formats. - Download models: Download discovered models. - Snoop: Monitor notebook execution and results until timeout. - Recurring jobs: Launch/Kill recurring snippets of code silently run in the target environment.
With pip install vger[ai] you'll get LLM generated summaries of notebooks in the target environment. These are meant to be rough translation for non-DS/AI folks to do quick triage of if (or which) notebooks are worth investigating further.
There was an inherent tradeoff on model size vs. ability and that's something I'll continue to tinker with, but hopefully this is helpful for some more traditional security users. I'd love to see folks start prompt injecting their notebooks ("these are not the droids you're looking for").
Microsoft today released updates to fix more than 60 security holes in Windows computers and supported software, including two “zero-day” vulnerabilities in Windows that are already being exploited in active attacks. There are also important security patches available for macOS and Adobe users, and for the Chrome Web browser, which just patched its own zero-day flaw.
First, the zero-days. CVE-2024-30051 is an “elevation of privilege” bug in a core Windows library. Satnam Narang at Tenable said this flaw is being used as part of post-compromise activity to elevate privileges as a local attacker.
“CVE-2024-30051 is used to gain initial access into a target environment and requires the use of social engineering tactics via email, social media or instant messaging to convince a target to open a specially crafted document file,” Narang said. “Once exploited, the attacker can bypass OLE mitigations in Microsoft 365 and Microsoft Office, which are security features designed to protect end users from malicious files.”
Kaspersky Lab, one of two companies credited with reporting exploitation of CVE-2024-30051 to Microsoft, has published a fascinating writeup on how they discovered the exploit in a file shared with Virustotal.com.
Kaspersky said it has since seen the exploit used together with QakBot and other malware. Emerging in 2007 as a banking trojan, QakBot (a.k.a. Qbot and Pinkslipbot) has morphed into an advanced malware strain now used by multiple cybercriminal groups to prepare newly compromised networks for ransomware infestations.
CVE-2024-30040 is a security feature bypass in MSHTML, a component that is deeply tied to the default Web browser on Windows systems. Microsoft’s advisory on this flaw is fairly sparse, but Kevin Breen from Immersive Labs said this vulnerability also affects Office 365 and Microsoft Office applications.
“Very little information is provided and the short description is painfully obtuse,” Breen said of Microsoft’s advisory on CVE-2024-30040.
The only vulnerability fixed this month that earned Microsoft’s most-dire “critical” rating is CVE-2024-30044, a flaw in Sharepoint that Microsoft said is likely to be exploited. Tenable’s Narang notes that exploitation of this bug requires an attacker to be authenticated to a vulnerable SharePoint Server with Site Owner permissions (or higher) first and to take additional steps in order to exploit this flaw, which makes this flaw less likely to be widely exploited as most attackers follow the path of least resistance.
Five days ago, Google released a security update for Chrome that fixes a zero-day in the popular browser. Chrome usually auto-downloads any available updates, but it still may require a complete restart of the browser to install them. If you use Chrome and see a “Relaunch to update” message in the upper right corner of the browser, it’s time to restart.
Apple has just shipped macOS Sonoma 14.5 update, which includes nearly two dozen security patches. To ensure your Mac is up-to-date, go to System Settings, General tab, then Software Update and follow any prompts.
Finally, Adobe has critical security patches available for a range of products, including Acrobat, Reader, Illustrator, Adobe Substance 3D Painter, Adobe Aero, Adobe Animate and Adobe Framemaker.
Regardless of whether you use a Mac or Windows system (or something else), it’s always a good idea to backup your data and or system before applying any security updates. For a closer look at the individual fixes released by Microsoft today, check out the complete list over at the SANS Internet Storm Center. Anyone in charge of maintaining Windows systems in an enterprise environment should keep an eye on askwoody.com, which usually has the scoop on any wonky Windows patches.
Update, May 15, 8:28 a.m.: Corrected misattribution of CVE-2024-30051.
There has been an exponential increase in breaches within enterprises despite the carefully constructed and controlled perimeters that exist around applications and data. Once an attacker can access… Read more on Cisco Blogs
Article tags
The full explanation what is HTML Smuggling may be found here.
The primary objective of HTML smuggling is to bypass network security controls, such as firewalls and intrusion detection systems, by disguising malicious payloads within seemingly harmless HTML and JavaScript code. By exploiting the dynamic nature of web applications, attackers can deliver malicious content to a user's browser without triggering security alerts or being detected by traditional security mechanisms. Thanks to this technique, the download of a malicious file is not displayed in any way in modern IDS solutions.
The main goal of HTMLSmuggler tool is creating an independent javascript library with embedded malicious user-defined payload. This library may be integrated into your phishing sites/email html attachments/etc. to bypass IDS and IPS system and deliver embedded payload to the target user system. An example of created javascript library may be found here.
Install yarn package manager.
Install dependencies:
yarnRead help message.
yarn build -hModify (or use my) javascript-obfuscator options in obfuscator.js, my preset is nice, but very slow.
Compile your javascript payload:
yarn build -p /path/to/payload -n file.exe -t "application/octet-stream" -cGet your payload from dist/payload.esm.js or dist/payload.umd.js. After that, it may be inserted into your page and called with download() function.
payload.esm.jsis used inimport { download } from 'payload.esm';imports (ECMAScript standart).
payload.umd.jsis used in html script SRC andrequire('payload.umd');imports (CommonJS, AMD and pure html).
A full example may be found here.
Import created script to html file (or insert it inline):
<head>
<script src="payload.umd.js"></script>
</head>Call download() function from body:
<body>
<button onclick="download()">Some phishy button</button>
</body>Happy phishing :)
A full example may be found here.
Import created script to vue file:
<script>
import { download } from './payload.esm';
</script>Call download() function:
<template>
<button @click="download()">Some phishy button</button>
</template>Happy phishing :)
Q: I have an error RangeError: Maximum call stack size exceeded, how to solve it?
A: This issue described here. To fix it, try to disable splitStrings in obfuscator.js or make smaller payload (it's recommended to use up to 2 MB payloads because of this issue).
Q: Why does my payload build so long?
A: The bigger payload you use, the longer it takes to create a JS file. To decrease time of build, try to disable splitStrings in obfuscator.js. Below is a table with estimated build times using default obfuscator.js.
| Payload size | Build time |
|---|---|
| 525 KB | 53 s |
| 1.25 MB | 8 m |
| 3.59 MB | 25 m |
The dystopian 2020s, ’30s, and ’40s depicted in novels and movies written and produced decades ago blessedly seem very far off from the timeline of reality. Yes, we have refrigerators that suggest grocery lists, and we have devices in our pockets that control seemingly every function of our homes. But there aren’t giant robots roaming the streets bellowing orders.
Humans are still very much in control of society. To keep it that way, we must use the latest technological advancements for their main intended purpose: To boost convenience in our busy lives.
The future of technology is bright. With the right attitude, security tools, and a reminder every now and then to look up from our devices, humans will be able to enjoy everything the future of technology holds.
A look into the future of technology would be incomplete without touching on the applications and impact of artificial intelligence (AI) on everyday tasks. Platforms like ChatGPT , Voice.ai, and Craiyon have thrilled, shocked, and unnerved the world in equal measures. AI has already transformed work life, home life, and free time of everyday people everywhere.
According to McAfee’s Modern Love Research Report, 26% of people would use AI to aid in composing a love note. Plus, more than two-thirds of those surveyed couldn’t tell the difference between a love note written by AI and a human. AI can be a good tool to generate ideas, but replacing genuine human emotion with the words of a computer program could create a shaky foundation for a relationship.
The Center for AI Safety urges that humans must take an active role in using AI responsibly. Cybercriminals and unsavory online characters are already using it maliciously to gain financially and spread incendiary misinformation. For example, AI-generated voice imposters are scamming concerned family members and friends with heartfelt pleas for financial help with a voice that sounds just like their loved one. Voice scams are turning out to be fruitful for scammers: 77% of people polled who received a cloned voice scam lost money as a result.
Even people who aren’t intending mischief can cause a considerable amount when they use AI to cut corners. One lawyer’s testimony went awry when his research partner, ChatGPT, when rogue and completely made up its justification.1 This phenomenon is known as an AI hallucination. It occurs when ChatGPT or other similar AI content generation tool doesn’t know the answer to your question, so it fabricates sources and asserts you that it’s giving you the truth.
Overreliance on ChatGPT’s output and immediately trusting it as truth can lead to an internet rampant with fake news and false accounts. Keep in mind that using ChatGPT introduces risk in the content creation process. Use it responsibly.
Though it’s powered by AI and could fall under the AI section above, deepfake is exploding and deserves its own spotlight. Deepfake technology is the manipulation of videos to digitally transform one person’s appearance resemble someone else, usually a public figure. Deepfake videos are often accompanied by AI-altered voice tracks. Deepfake challenges the accuracy of the common saying, “Seeing is believing.” Now, it’s more difficult than ever to separate fact from fiction.
Not all deepfake uses are nefarious. Deepfake could become a valuable tool in special effects and editing for the gaming and film industries. Additionally, fugitive sketch artists could leverage deepfake to create ultra-realistic portraits of wanted criminals. If you decide to use deepfake to add some flair to your social media feed or portfolio, make sure to add a disclaimer that you altered reality with the technology.
Currently, it’s estimated that there are more than 15 billion connected devices in the world. A connected device is defined as anything that connects to the internet. In addition to smartphones, computers, and tablets, connected devices also extend to smart refrigerators, smart lightbulbs, smart TVs, virtual home assistants, smart thermostats, etc. By 2030, there may be as many as 29 billion connected devices.2
The growing number of connected devices can be attributed to our desire for convenience. The ability to remote start your car on a frigid morning from the comfort of your home would’ve been a dream in the 1990s. Checking your refrigerator’s contents from the grocery store eliminates the need for a pesky second trip to pick up the items you forgot the first time around.
The downfall of so many connected devices is that it presents crybercriminals literally billions of opportunities to steal people’s personally identifiable information. Each device is a window into your online life, so it’s essential to guard each device to keep cybercriminals away from your important personal details.
With the widespread adoption of email, then cellphones, and then social media in the ’80s, ’90s and early 2000s, respectively, people have integrated technology into their daily lives that better helps them connect with other people. More recent technological innovations seem to trend toward how to connect people to their other devices for a seamless digital life.
We shouldn’t ignore that the more devices and online accounts we manage, the more opportunities cybercriminals have to weasel their way into your digital life and put your personally identifiable information at risk. To protect your online privacy, devices, and identity, entrust your digital safety to McAfee+. McAfee+ includes $1 million in identity theft coverage, virtual private network (VPN), Personal Data Cleanup, and more.
The future isn’t a scary place. It’s a place of infinite technological possibilities! Explore them confidently with McAfee+ by your side.
1The New York Times, “Here’s What Happens When Your Lawyer Uses ChatGPT”
2Statista, “Number of Internet of Things (IoT) connected devices worldwide from 2019 to 2021, with forecasts from 2022 to 2030”
The post The Future of Technology: AI, Deepfake, & Connected Devices appeared first on McAfee Blog.
A DLL Loader With Advanced Evasive Features
"Atom" function via the command line.CRC32 string hashing algorithm.AtomLdr's unhooking method looks like the following
the program Unhooking from the \KnwonDlls\ directory is not a new method to bypass user-land hooks. However, this loader tries to avoid allocating RWX memory when doing so. This was obligatory to do in KnownDllUnhook for example, where RWX permissions were needed to replace the text section of the hooked modules, and at the same time allow execution of functions within these text sections.
This was changed in this loader, where it suspends the running threads, in an attempt to block any function from being called from within the targetted text sections, thus eliminating the need of having them marked as RWX sections before unhooking, making RW permissions a possible choice.
This approach, however, created another problem; when unhooking, NtProtectVirtualMemory syscall and others were using the syscall instruction inside of ntdll.dll module, as an indirect-syscall approach. Still, as mentioned above, the unhooked modules will be marked as RW sections, making it impossible to perform indirect syscalls, because the syscall instruction that we were jumping to, can't be executed now, so we had to jump to another executable place, this is where win32u.dll was used.
win32u.dll contains some syscalls that are GUI-related functions, making it suitable to jump to instead of ntdll.dll. win32u.dll is loaded (statically), but not included in the unhooking routine, which is done to insure that win32u.dll can still execute the syscall instruction we are jumping to.
The suspended threads after that are resumed.
It is worth mentioning that this approach may not be that efficient, and can be unstable, that is due to the thread suspension trick used. However, it has been tested with multiple processes with positive results, in the meantime, if you encountered any problems, feel free to open an issue.
PayloadConfig.pc file, that contains the encrypted payload, and its encrypted key and iv.PayloadConfig.pc file will then replace this in the AtomLdr project.AtomLdr project as x64 Release.AtomLdr.dll using rundll32.exe, running Havoc payload, and capturing a screenshotAtomLdr.dll's Import Address TablePayloadBuilder.exe, to encrypt demon[111].bin - a Havoc payload fileAtomLdr.dll using rundll32.exe
CMLoot was created to easily find interesting files stored on System Center Configuration Manager (SCCM/CM) SMB shares. The shares are used for distributing software to Windows clients in Windows enterprise environments and can contains scripts/configuration files with passwords, certificates (pfx), etc. Most SCCM deployments are configured to allow all users to read the files on the shares, sometimes it is limited to computer accounts.
The Content Library of SCCM/CM have a "complex" (annoying) file structure which CMLoot will untangle for you: https://techcommunity.microsoft.com/t5/configuration-manager-archive/understanding-the-configuration-manager-content-library/ba-p/273349
Essentially the DataLib folder contains .INI files, the .INI file are named the original filename + .INI. The .INI file contains a hash of the file, and the file itself is stored in the FileLib in format of <folder name: 4 first chars of the hash>\fullhash.
It is possible to apply Access control to packages in CM. This however only protects the folder for the file descriptor (DataLib), not the actual file itself. CMLoot will during inventory record any package that it can't access (Access denied) to the file _noaccess.txt. Invoke-CMLootHunt can then use this file to enumerate the actual files that the access control is trying to protect.
Windows Defender for Endpoint (EDR) or other security mechanisms might trigger because the script parses a lot of files over SMB.
Find CM servers by searching for them in Active Directory or by fetching this reqistry key on a workstation with System Center installed:
(Get-ItemProperty -Path HKLM:\SOFTWARE\Microsoft\SMS\DP -Name ManagementPoints).ManagementPoints
There may be multiple CM servers deployed and they can contain different files so be sure to find all of them.
Then you need to create an inventory file which is just a text file containing references to file descriptors (.INI). The following command will parse all .INI files on the SCCM server to create a list of files available.
PS> Invoke-CMLootInventory -SCCMHost sccm01.domain.local -Outfile sccmfiles.txt
Then use the inventory file created above to download files of interest:
Select files using GridView (Milage may vary with large inventory files):
PS> Invoke-CMLootDownload -InventoryFile .\sccmfiles.txt -GridSelect
Download a single file, by coping a line in the inventory text:
PS> Invoke-CMLootDownload -SingleFile \\sccm\SCCMContentLib$\DataLib\SC100001.1\x86\MigApp.xml
Download all files with a certain file extension:
PS> Invoke-CMLootDownload -InventoryFile .\sccmfiles.txt -Extension ps1
Files will by default download to CMLootOut in the folder from which you execute the script, can be changed with -OutFolder parameter. Files are saved in the format of (folder: filext)\(first 4 chars of hash>_original filename).
Hunt for files that CMLootInventory found inaccessible:
Invoke-CMLootHunt -SCCMHost sccm -NoAccessFile sccmfiles_noaccess.txt
Bulk extract MSI files:
Invoke-CMLootExtract -Path .\CMLootOut\msi
Run inventory, scanning available files:
Select files using GridSelect:
Hunt "inaccessible" files and MSI extract:
Tomas Rzepka / WithSecure
Photo: BrandonKleinPhoto / Shutterstock.com
Two U.S. men have been charged with hacking into the Ring home security cameras of a dozen random people and then “swatting” them — falsely reporting a violent incident at the target’s address to trick local police into responding with force. Prosecutors say the duo used the compromised Ring devices to stream live video footage on social media of police raiding their targets’ homes, and to taunt authorities when they arrived.
Prosecutors in Los Angeles allege 20-year-old James Thomas Andrew McCarty, a.k.a. “Aspertaine,” of Charlotte, N.C., and Kya Christian Nelson, a.k.a. “ChumLul,” 22, of Racine, Wisc., conspired to hack into Yahoo email accounts belonging to victims in the United States. From there, the two allegedly would check how many of those Yahoo accounts were associated with Ring accounts, and then target people who used the same password for both accounts.
An indictment unsealed this week says that in the span of just one week in November 2020, McCarty and Nelson identified and swatted at least a dozen different victims across the country.
“The defendants then allegedly accessed without authorization the victims’ Ring devices and transmitted the audio and video from those devices on social media during the police response,” reads a statement from Martin Estrada, the U.S. Attorney for the Central District of California. “They also allegedly verbally taunted responding police officers and victims through the Ring devices during several of the incidents.”
James Thomas Andrew McCarty.
The indictment charges that McCarty continued his swatting spree in 2021 from his hometown in Kayenta, Ariz., where he called in bomb threats or phony hostage situations on more than two dozen occasions.
The Telegram and Discord aliases allegedly used by McCarty — “Aspertaine” and “Couch,” among others — correspond to an identity that was active in certain channels dedicated to SIM-swapping, a crime that involves stealing wireless phone numbers and hijacking the online financial and social media accounts tied to those numbers.
Aspertaine bragged on Discord that he’d amassed more than $330,000 in virtual currency. On Telegram, the Aspertaine/Couch alias frequented several popular SIM-swapping channels, where they initially were active as a “holder” — a SIM-swapping group member who agrees to hold SIM cards used in the heist after an account takeover is completed. Aspertaine later claimed more direct involvement in individual SIM-swapping attacks.
In September, KrebsOnSecurity broke the news about a wide-ranging federal investigation into “violence-as-a-service” offerings on Telegram and other social media networks, wherein people can settle scores by hiring total strangers to carry out physical attacks such as brickings, shootings, and firebombings at a target’s address.
The story observed that SIM swappers were especially enamored of these “IRL” or “In Real Life” violence services, which they frequently used to target one another in response to disagreements over how stolen money should be divided amongst themselves. And a number of Aspertaine’s peers on these SIM-swapping channels claimed they’d been ripped off after Aspertaine took more than a fair share from them.
In August, a member of a popular SIM-swapping group on Telegram who was slighted by Aspertaine put out the word that he was looking for some physical violence to be visited on McCarty’s address in North Carolina. “Anyone live near here and wants to [do] a job for me,” the job ad with McCarty’s home address read. “Jobs range from $1k-$50k. Payment in BTC [bitcoin].” It’s unclear if anyone responded to that job offer.
Ring, Inc., which is owned by Amazon, said it learned bad actors used stolen customer email credentials obtained from external (non-Ring) services to access other accounts, and took immediate steps to help those customers secure their Ring accounts.
“We also supported the FBI in identifying the individuals responsible,” the company said in a written statement. “We take the security of our customers extremely seriously — that’s why we made two-step verification mandatory, conduct regular scans for Ring passwords compromised in non-Ring breaches, and continually invest in new security protections to harden our systems. We are committed to continuing to protect our customers and vigorously going after those who seek to harm them.”
KrebsOnSecurity recently published The Wages of Password ReUse: Your Money or Your Life, which noted that when normal computer users fall into the nasty habit of recycling passwords, the result is most often some type of financial loss. Whereas, when cybercriminals reuse passwords, it often costs them their freedom.
But perhaps that story should be updated, because it’s now clear that password reuse can also put you in mortal danger. Swatting attacks are dangerous, expensive hoaxes that sometimes end in tragedy.
In June 2021, an 18-year-old serial swatter from Tennessee was sentenced to five years in prison for his role in a fraudulent swatting attack that led to the death of a 60-year-old man.
In 2019, prosecutors handed down a 20-year sentence to Tyler Barriss, a then 26-year-old serial swatter from California who admitted making a phony emergency call to police in late 2017 that led to the shooting death of an innocent Kansas man.
McCarty was arrested last week, and charged with conspiracy to intentionally access computers without authorization. Prosecutors said Nelson is currently incarcerated in Kentucky in connection with unrelated investigation.
If convicted on the conspiracy charge, both defendants would face a statutory maximum penalty of five years in federal prison. The charge of intentionally accessing without authorization a computer carries a maximum possible sentence of five years. A conviction on the additional charge against Nelson — aggravated identity theft — carries a mandatory two-year consecutive sentence.
Update, 11:48 a.m., Dec. 20: Added statement from Ring. Modified description of a “holder” in the SIM-swapping parlance.
AMLBot, a service that helps businesses avoid transacting with cryptocurrency wallets that have been sanctioned for cybercrime activity, said an investigation published by KrebsOnSecurity last year helped it shut down three dark web services that secretly resold its technology to help cybercrooks avoid detection by anti-money laundering systems.
Antinalysis, as it existed in 2021.
In August 2021, KrebsOnSecurity published “New Anti Anti-Money Laundering Services for Crooks,” which examined Antinalysis, a service marketed on cybercrime forums that purported to offer a glimpse of how one’s payment activity might be flagged by law enforcement agencies and private companies that track and trace cryptocurrency transactions.
“Worried about dirty funds in your BTC address? Come check out Antinalysis, the new address risk analyzer,” read the service’s opening announcement. “This service is dedicated to individuals that have the need to possess complete privacy on the blockchain, offering a perspective from the opponent’s point of view in order for the user to comprehend the possibility of his/her funds getting flagged down under autocratic illegal charges.”
Antinalysis allows free lookups, but anyone wishing to conduct bulk look-ups has to pay at least USD $3, with a minimum $30 purchase. Other plans go for as high as $6,000 for 5,000 requests. Nick Bax, a security researcher who specializes in tracing cryptocurrency transactions, told KrebsOnSecurity at the time that Antinalysis was likely a clone of AMLBot because the two services generated near-identical results.
AMLBot shut down Antinalysis’s access just hours after last year’s story went live. However, Antinalysis[.]org remains online and accepting requests, as does the service’s Tor-based domain, and it is unclear how those services are sourcing their information.
AMLBot spokesperson Polina Smoliar said the company undertook a thorough review after that discovery, and in the process found two other services similar to Antinalysis that were reselling their application programming interface (API) access to cybercrooks.
Smoliar said that following the revelations about Antinalysis, AMLBot audited its entire client base, and implemented the ability to provide APIs only after a contract is signed and the client has been fully audited. AMLBot said it also instituted 24/7 monitoring of all client transactions.
“As a result of these actions, two more services with the name AML (the same as AMLBot has) were found to be involved in fraudulent schemes,” Smoliar said. “Information about the fraudsters was also sent to key market participants, and their transaction data was added to the tracking database to better combat money laundering.”
Experts say the founder of Antinalysis also runs a darknet market for narcotics.
The Antinalysis homepage and chatter on the cybercrime forums indicates the service was created by a group of coders known as the Incognito Team. Tom Robinson, co-founder of the blockchain intelligence firm Elliptic, said the creator of Antinalysis is also one of the developers of Incognito Market, a darknet marketplace specializing in the sale of narcotics.
“Incognito was launched in late 2020, and accepts payments in both Bitcoin and Monero, a cryptoasset offering heightened anonymity,” Robinson said. “The launch of Antinalysis likely reflects the difficulties faced by the market and its vendors in cashing out their Bitcoin proceeds.”
Chief Information Officers and other technology decision makers continuously seek new and better ways to evaluate and manage their investments in innovation – especially the technologies that may create consequential decisions that impact human rights. As Artificial Intelligence (AI) becomes more prominent in vendor offerings, there is an increasing need to identify, manage, and mitigate the unique risks that AI-based technologies may bring.
Cisco is committed to maintaining a responsible, fair, and reflective approach to the governance, implementation, and use of AI technologies in our solutions. The Cisco Responsible AI initiative maximizes the potential benefits of AI while mitigating bias or inappropriate use of these technologies.
Gartner® Research recently published “Innovation Insight for Bias Detection/Mitigation, Explainable AI and Interpretable AI,” offering guidance on the best ways to incorporate AI-based solutions that facilitates “understanding, trust and performance accountability required by stakeholders.” This newsletter describes Cisco’s approach to Responsible AI governance and features this Gartner report.
At Cisco, we are committed to managing AI development in a way that augments our focus on security, privacy, and human rights. The Cisco Responsible AI initiative and framework governs the application of responsible AI controls in our product development lifecycle, how we manage incidents that arise, engage externally, and its use across Cisco’s solutions, services, and enterprise operations.
Our Responsible AI framework comprises:
We base our Responsible AI initiative on principles consistent with Cisco’s operating practices and directly applicable to the governance of AI innovation. These principles—Transparency, Fairness, Accountability, Privacy, Security, and Reliability—are used to upskill our development teams to map to controls in the Cisco Secure Development Lifecycle and embed Security by Design, Privacy by Design, and Human Rights by Design in our solutions. And our principle-based approach empowers customers to take part in a continuous feedback cycle that informs our development process.
We strive to meet the highest standards of these principles when developing, deploying, and operating AI-based solutions to respect human rights, encourage innovation, and serve Cisco’s purpose to power an inclusive future for all.
Check out Gartner recommendations for integrating AI into an organization’s data systems in this Newsletter and learn more about Cisco’s approach to Responsible Innovation by reading our introduction “Transparency Is Key: Introducing Cisco Responsible AI.”
We’d love to hear what you think. Ask a Question, Comment Below, and Stay Connected with Cisco Secure on social!
Cisco Secure Social Channels
