Robust Intelligence
@robusthq
Followers
1,988
Following
68
Media
93
Statuses
277
Achieve AI security and safety to unblock the enterprise AI mission.
San Francisco, CA
Joined August 2020
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
#GFRunTheWorldConcertD1
• 415418 Tweets
Napoli
• 136264 Tweets
Ticketmaster
• 64120 Tweets
KINGPOWER x PHUWIN
• 56251 Tweets
#خلصوا_صفقات_الهلال2
• 40445 Tweets
#PVL2024
• 39973 Tweets
Akari
• 30188 Tweets
Qoo10
• 29925 Tweets
DONBELLE DAOG SA DAVAO
• 24599 Tweets
ポケミク
• 20301 Tweets
#Oasis25
• 18732 Tweets
ポムの樹
• 18389 Tweets
PLDT
• 17696 Tweets
月ノ美兎
• 15966 Tweets
椎名唯華
• 15727 Tweets
YOSHIKI
• 12202 Tweets
#ドッキリGP
• 12155 Tweets
We're excited to share that
@aminkarbasi
has joined RI as our Chief Science Officer, where he will lead our AI security research agenda & initiatives.
We look forward to all we’ll accomplish together on our mission to enable every organization on the planet to adopt AI securely!
10
3
67
It’s no secret today’s machine learning models fail frequently, which can have dire consequences when used to make critical decisions. This is why companies are actively building strategies and engineering paradigms to instill ML integrity in their systems.
1
7
13
We're proud to announce a technology partnership with
@databricks
to deliver ML integrity through continuous testing!
Read our blog to learn how teams can seamlessly scale their MLOps pipeline while ensuring trust in their models.
0
3
13
We're proud to co-author an update to
@NIST
's "Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations" report, which includes security of LLMs.
This report is the result of incredible work by
@ApostolVassilev
@AlinaMOprea
@alieJfordyce
@drhyrum
👏
Adversaries can deliberately manipulate AI systems to make them malfunction — and there’s no foolproof defense.
Check out NIST’s new guide to the types of attacks developers and users can expect, along with approaches to mitigate them:
7
107
223
1
4
13
Within hours of OpenAI's release of Structured Outputs, our AI security researchers identified a simple yet concerning exploit that bypasses the model's safety measures, including its refusal capabilities.
See our blog for in-depth analysis:
2
2
15
🚨 Join us for a talk on adversarial ML.
@drhyrum
will host fellow co-authors
@g_apru
,
@fbpierazzi
& David Freeman to discuss their recent paper, which was accepted to IEEE SaTML23.
Register below:
*Credit also to co-authors
@savinodam
&
@AlejandroRoundy
1
7
13
📢 Announcing a new generative AI risk assessment report on Llama-2-70B. Check out the report to learn about our testing methodology and the security, ethical & operational risks we identified.
Read it:
#airisk
#airiskmanagement
#llmsecurity
0
2
7
We’re excited to announce we’ve closed a $30M Series B to lead the future of eliminating AI risk 🚀 🚀 🚀 Read more about how we plan to do this in our recent interview with
@fredericl
for
@TechCrunch
:
0
4
12
We'd like to celebrate our colleague Blaine Nelson, as well as co-authors
@biggiobattista
and Pavel Laskov, on receiving the
@icmlconf
's prestigious Test of Time award yesterday! Their paper "Poisoning Attacks Against Support Vector Machines" has indeed had lasting impact!
1
3
11
We're proud to announce our Series A and look forward to keeping you appraised of our efforts to defeat adversarial AI attacks.
1
4
11
New to adversarial machine learning?
@lexfridman
's interview with
@dawnsongtweets
is a great way to start learning about the field.
0
6
9
🏆 Robust Intelligence is recognized as the "Best AI Startup" by AI Breakthrough Awards!
Thank you to our incredible team for your drive and passion, as well as our customers for your partnership. This award is for you!
Check out the press release below:
0
4
10
📺
@robusthq
co-founder
@kojinoshiba
was interviewed about navigating the risks of generative AI alongside
@AndrewYNg
,
@mer__edith
& Yoshua Bengio.
This expert panel discussed GAI's impact on business and society, new risks & impending regulation.
0
5
10
The new addition to the RI HQ! Barista station with a LED logo in progress!
#LoveWhereYouWork
#AI
#robustintelligence
0
1
9
In preliminary analysis of GPT-4 vs GPT-4o, we saw higher susceptibility to jailbreaks on GPT-4o with fewer queries. Higher performance doesn't mean safer models!
Learn more about algorithmic jailbreaking and AI security measures:
2
3
10
🏆 We're honored to receive the "Best AI Startup" award for the second consecutive year from
@AI_Breakthrough
!
Chosen from over 5,000 global nominations, this award highlights our commitment to innovation in AI application security.
0
1
9
Model safety is just one important component of AI risk management - an essential building block of
#GenerativeAI
. Read the article below to learn the role plays in mitigating the security, ethical, and operational risks of AI.
#airisk
#aisecurity
#llms
0
2
9
Today, we are proud to announce our partnership with Tokio Marine, one of the largest insurance groups in Japan! Together, we will promote AI security to the Japanese market and the insurance industry globally.
0
2
8
AI threat research is the foundation of our platform and a practice that helps us keep our customers protected as tech evolves and new vulnerabilities emerge.
We’re pleased to share our latest monthly threat roundup with the
#AIsecurity
community!
AI Cyber Threat Intelligence Roundup: February 2024 — Robust Intelligence
www.robustintelligence.com
0
2
4
Have you considered red teaming your
#AI
systems? Earlier this week, Robust Intelligence Distinguished Engineer
@drhyrum
explored the pros and cons of this very topic with fellow panelists from
@GoogleAI
&
@MITREcorp
. RSA Conference passholders can watch the discussion on-demand.
1
3
8
"Mrs. Kadelle explained that a good lifeguard should never have to enter the water to make a rescue. Instead, a good lifeguard will prevent their patrons from ever needing to be rescued, by continually monitoring the pool."
Can you see the analogy? 🏊👇
0
3
8
Our CEO,
@yaronsinger
, will be giving a talk at
@MLOpsWorld
Annual Summit tomorrow (06/09): “Eliminating AI Risk, One Model Failure at a Time”. If you’re in attendance, stop by Grand West to see it live and follow-up with Yaron at booth 16!
0
3
6
Security leaders are beginning to address the "unmanaged risk" of GenAI, which is critical to the safe use of apps built on these models.
Forrester suggests prioritizing AI security & recognizes that new tactics are required to mitigate risk.
What's your AI security strategy?
2
1
6
Want to know the best practices to eliminate AI risk in your company? Join
@yaronsinger
's talk tomorrow morning at the Databricks
#DataAISummit
. This hybrid session will be delivered live from Moscone South and streamed free online. Hope to see you there!
0
1
7
Thrilled to be included in a16z + Future's Data50 -- the World's Top 50 Data Startups List 🚀🚀🚀
0
4
7
🏆 We're thrilled to announce that
@FastCompany
named us the Most Innovative Data Science Company of 2023!
We’re excited for what’s to come. Come join us on our journey to create machine learning integrity and mitigate
#AI
risk:
#FCMostInnovative
0
2
7
The 2nd entry in our AI risk analysis series features Dolly 2.0 — an experimental LLM open-sourced by Databricks in March 2023. We explore risks that are presented by typical tasks, example code, and more.
Read more on our blog
#LLM
#GenerativeAI
#AIRisk
0
2
7
Great article in WIRED about the risk of indirect prompt injection attacks on
#generativeAI
, featuring our work at
@robusthq
.
Thanks
@mattburgess1
for brining attention to this pressing AI security issue in a way that's both accessible and educational.
1
3
6
New video from
@Stanford
MLSys Seminar
@yaronsinger
explains the root causes for misbehaving AI and presents a rigorous framework for building truly robust AI.
0
3
6
📣 We’re thrilled to announce our partnership with
@IBMwatsonx
, making it easy to protect all
#GenAI
models and applications on watsonx AI from safety and security threats in real time!
Below, see how to route all calls through our AI Firewall in just a few lines of code.
0
2
6
AI risk has become a pressing concern for companies, but most don't have a mitigation strategy. Join us for a webinar to discuss practical solutions to security, ethical, and operational AI risks.
Space is limited to industry leaders. Register below:
0
2
4
Check out our latest
@Medium
post about exposing vulnerabilities in check processing systems:
#adversarialmachinelearning
#AIsecurity
0
2
6
Curious to learn how executives at leading companies instill integrity in their ML systems?
Hear about the strategies and engineering paradigms used to eliminate model failure at Equinix, JPMorgan, Splunk, Survey Monkey, Wells Fargo and more!
Register for ML:Integrity on Oct 19
Join industry leaders at the ML:Integrity conference!
It's a free, virtual conference with speakers from
@awscloud
,
@huggingface
,
@DeepMind
,
@Reddit
,
@PayPal
and more!
Register for free here:
0
2
10
0
1
6
Our CTO
@drhyrum
be joining
@Unusual_VC
VP
@allisonaverill
on 11/16 for a fireside chat. They will discuss threats to AI systems, how the AI attack surfacing is evolving, and how enterprises can secure their systems. Register:
#airiskmanagement
#aisecurity
0
1
6
President Biden issued a highly anticipated executive order that establishes new actions on the safe, secure, and trustworthy use of
#AI
.
While there are many parts to the order, AI red teaming is a central component.
Check out our blog to learn more!
1
2
6
Harrison Chase, a Lead MLE at Robust Intelligence, explores malformed inputs and how common machine learning packages handle them.
The results were surprising! Different ML packages produce inconsistent behavior when given the same malformed input.
👇👇
0
8
6
Check out our latest blog post about production-ready AI models!
1
0
6
Join us at
@BlueHatIL
to learn how to address the unmanaged security risk posed by AI.
@drhyrum
will share insights about current AI vulnerabilities and the key role that security practitioners must play to ensure the safe use of chatbots, GPT-4 & more.
0
2
6
Robust Intelligence is honored to be selected as "MLOps Company of the Year" by
@AI_Breakthrough
in their fifth annual awards program. 🏆
Read the joint press release here 👇
#ai
#machinelearning
#mlops
1
2
6
A great conversation between
@jeremiecharris
and
@yaronsinger
- thanks for having us
@TDataScience
!
How can we build AI systems that are more robust, safer, and savvier at tackling faulty data? In our latest TDS Podcast episode, host
@jeremiecharris
chatted about these (and other) topics with
@robusthq
CEO
@yaronsinger
.
1
2
10
0
3
6
"With issues like bugs and broken data pipelines, your AI system will crash, literally.
You have to firefight these issues in production.
You waste precious time digging through error logs, identifying the root cause, all while your model is crashing."
🥵 Sound familiar? 👇
1
3
5
Attending the
@ai_expo
in Silicon Valley next week? So are we!
@kojinoshiba
is speaking on the 'Keeping it Ethical in AI' panel.
We'll be at booth 34. Stop by to say hi, and ask us how we help companies eliminate model failure through ML integrity.
#aiandbigdataexpo
0
1
5
What is algorithmic AI
#redteaming
and how does it impact the security of your GenAI apps?
Check out our new interactive guide to learn about this automated prompt injection technique capable of jailbreaking sophisticated
#LLMs
with no human supervision.
0
1
5
Congrats to Robust Intelligence co-founder
@kojinoshiba
on his selection to
#ForbesUnder30
!
Kojin, it's truly a privilege to work with you. Thank you for your vision, leadership & contributions to the advancement of
#AIsecurity
!
Read his Forbes profile:
0
0
5
Announcing the
@AIRiskDatabase
, a free and community-supported resource to evaluate AI supply chain risk in open source models.
Check out the AI Risk Database:
Visit our blog for initial findings on vulnerabilities:
0
3
5
In our latest blog post, ML engineer
@aufieroma12
shares the security risks of generative AI OSS.
See the link below to learn about the emerging security challenges & recommendations to deploy AI systems safely.
#adversarialML
#generativeai
#opensource
0
0
5
📣 We're thrilled to announce our AI security partnership with
@MongoDB
, combining the power of AI Firewall and Atlas Vector Search!
Check out our latest blog post, where you can sign-up for AI Firewall today.
#AIsecurity
#generativeAI
#LLMs
#guardrails
0
1
5
📣Join us on 9/21 at 11am PT to learn about AI red teaming!
@drhyrum
and
@FionnualaHowell
will discuss how to red team AI systems, why internal and external testing is important, and what AI red teaming looks like in the real world.
#AIsecurity
#AIrisk
0
1
5
2 adversarial ML experts walk into a room... join
@drhyrum
&
@AlejandroRoundy
at
#RSAC2023
to learn real world examples of ML anti-phishing detector evasion!
For more info, visit the RSA conference agenda:
1
0
5
As more corporations and government agencies recognize the threat posed by adversarial machine learning attacks, AI redteaming has become an important defense mechanism.
0
0
5
We've partnered with
@MITREcorp
to tackle AI supply chain risk in open-source models! Read more below.
If you're at
@BlackHatEvents
, swing by our joint Arsenal Lab sessions to see the enhanced AI Risk Database in action:
Aug 9 @ 12:15pm & Aug 10 @ 10am (Business Hall)
0
1
5
Last week, cybersecurity experts and Congressional staffers came together for a generative red team (GRT) exercise and panel discussions.
@robusthq
was both a participant and planning committee member.
For more, read the
@axios
article by
@samsabin923
:
0
1
4
⚠️ We identified an exploit in Meta's Prompt Guard - the mere act of adding spaces and removing punctuation resulted in a near 100% evasion rate for prompt injection.
We filed an issue and offered countermeasures to Meta.
📰 Article by
@ThomasClaburn
:
0
3
5
Excited to be included on the
#AI100
list - thanks
@CBinsights
!
Grateful for the validation for our work so far, but we're just getting started.
We're hiring across all our teams - come and join us to help build a secure future for AI.
0
0
5
Adversarial attacks can target all stages of the machine learning pipeline, including seemingly safe preprocessing steps like scaling images to the appropriate input dimensions.
0
0
5
We’re honored to be recognized by
@Rakuten
with an award for technology excellence.
Thank you Rakuten! It’s a privilege to be selected by a company with such high technology standards and bold vision.
#MLIntegrity
#AI
#MachineLearing
#DataScience
#Awards
#Rakuten
1
0
5
Check out our newest blog post on Subset Robustness in NLP!
"In this first part of our series on Robustness in NLP, we will discuss subpopulation performance and how subset testing can guide you in shipping a quality language-enabled product."
0
0
5
Check out Robust Intelligence's July 4th 🇺🇸 themed blog post!
If you like
#NLP
, historical themed
#machinelearning
experiments, or are just feeling particularly patriotic; this is a must-read!
Happy 4th 🇺🇸 🇺🇸 🇺🇸
0
2
5
A quick summary of where we are, and where we hope to go:
1
3
5
Meet Ben! Before joining RI over a year ago, Ben graduated from Stanford University with a degree in Computer Science. He brings enthusiasm and thought-provoking ideas to the office everyday.
#robustintelligence
#stanforduniversity
0
2
5
Learn practical advice for bringing
#LLMs
to production from experts
@nickpatience
, Nitin Wagh &
@kojin
.
They'll discuss ways to:
> test & validate models in dev
> protect models & data in prod
> comply with AI regulations, guidelines & frameworks
0
2
5
When hospitals suffer from a lack of specialists or radiologists, deep neural networks have been used to diagnose Covid-19 cases from chest x-rays. Recent research shows these models are susceptible to adversarial attacks, but defenses are possible.
0
0
4
Here's a video demonstrating how leading image classification models can be successfully attacked with our technology:
@KingJames
morphs into
@Drake
with minimal changes to an image.
1
2
4
Military commanders will need to trust that AI-enabled tools and systems are robust and secure before fully adopting them.
0
0
3
The AI security community is growing with every passing day! Robust Intelligence is proud to sponsor
@aivillage_dc
this year.
#defcon32
#defcon2024
This year we are hosting Generative Red Team - Part 2 (GRT2), talks from distinguished speakers, deepfake demo stations, and Dunk-a-Fed at the AI Village & Blue Team Village Pool Party. These events are only possible through the support of our sponsors
#thankyou
1
7
14
0
2
4
Insurance is on the cusp of a complete transformation at the hands of Artificial Intelligence.
Check out best practices from our partners Tokio Marine, who are leading the industry in building robust AI for insurance!
#AI
#insurance
0
5
4
In case you missed it... Read up on our Sales Operation Lead,
@DominicGlover10
!
"Being an athlete is undeniably an amazing foundation for this kind of job..."
#robustintelligence
#robustintelligencejobs
0
1
4
Robust Intelligence loves
#Sweetgreen
for lunch! Started using Sweetgreen's outpost.
#healthy
#salads
#tech
0
1
4
Introducing the ML Model Attribution Challenge, a technical competition designed to spur creative approaches to identifying the true origin of fine-tuned LLMs.
The competition kicks off at
@AIVillage_dc
on August 12. We hope you join!
0
1
4
Adversarial patches, which appear as logos superimposed on an image, are one way that object detection models can be fooled out of making otherwise straightforward classifications.
0
0
4
We’re pleased to host this panel of
#AIsecurity
experts who have each made outsized contributions to the advancement of NIST, MITRE & OWASP standards.
We hope you can join us! Special thanks to
@ApostolVassilev
,
@CLiaghati
^
@robvanderveer
.
Register here
0
1
4
Join us for our talk on testing
#generativeAI
models at the
@databricks
Data+AI Summit.
Robust Intelligence is a proud sponsor of the event. Visit our booth to learn how you can protect against AI risk through continuous validation within Databricks.
#DataAISummit
#LLMs
#AIrisk
0
2
4
Our CEO
@yaronsinger
will be speaking at LLM Avalanche on
#aisecurity
this coming Monday 6/26
If you weren't able to get tickets to the sold-out event, we'd be happy to connect at
#dataaisummit
. Find us at booth # 24 in the expo hall!
0
0
4
Check out
@yaronsinger
's conversation with the Practical AI team on Eliminating AI Failures below 👇👇👇
🚀 New episode of Practical AI! 🚀
🗣 Eliminate AI failures
🤩 with
@yaronsinger
from
@robusthq
🎙 hosts
@chrisbenson
@dwhitena
🗄️
#ai
#machinelearning
#datascience
💚
1
3
4
0
3
4
Attending the
#DataAISummit
next week? Don’t miss
@yaronsinger
's talk on eliminating AI failure. For those in SF, we invite you to request a private meeting with Yaron and our eng leadership. Use this link 👉
0
1
4
Don’t miss this
#rsac2023
talk!
@drhyrum
&
@AlejandroRoundy
will share how attackers evade ML anti-phishing detectors using use clever techniques.
Learn real world examples & measure the use of automated adversarial-attack creation by phishers.
0
0
4
Deepfakes are a growing part of online misinformation campaigns, but ongoing innovation seeks to identify and thwart their spread.
0
0
4
Deepfake audio and video will complicate collection and presentation of evidence in trials around the world. Detection of this adversarially generated content will be of paramount importance to the administration of justice.
1
0
4
Today, the White House announced new actions to further promote responsible AI innovation and protect people’s rights and safety.
The announcement includes 3 new initiatives.
#AI
#machinelearning
#AIintegrity
#responsibleAI
#AIpolicy
1
0
4
Historically, operating systems were deployed unprotected and defense came years later.
@goodfellow_ian
suggests we shouldn't go down the same route with machine learning systems, and we couldn't agree more.
1
0
4
Thanks to
@SVB_Financial
and In-Q-Tel for inviting us to participate in their Discovery Event today.
@yaronsinger
was happy to share insights on mitigating AI risk with the assembled audience of Fortune 500 CXOs, VCs, and government executives.
#ai
#ml
0
0
4
Ready to chat about AI security? Find us at RSA to join the conversation.
Reach out to schedule time with
@yaronsinger
and
@drhyrum
, and don't miss Hyrum's talk on Thursday about machine learning evasion.
#rsac2023
#AIintegrity
#AIrisk
#AIsecurity
#AI
#machinelearning
0
2
4
🕵What's on your short list for
#BlackHat
?
If your company is building
#GenAI
applications, meet us to learn about our
#AIsecurity
platform that is trusted by leading enterprises worldwide.
Schedule time with our leadership team:
0
2
3
Much adversarial machine learning research focuses on unstructured data like images and text. But tabular data, which may have complex validity rules and mixed feature types, are still susceptible to adversarial attacks.
0
0
2
We're excited to kick off our AI risk analysis series of popular 3rd-party models, featuring techniques similar to our automated, continuous validation solution. Read more on our blog here:
0
0
3
🕵️♂️ Securing your AI applications requires a novel
#cybersecurity
approach.
Schedule time to meet with us at
#RSAC
to learn how we help leading enterprises operationalize
#AIsecurity
policy to meet rigorous standards.
We look forward to meeting you!
0
0
3
"The 3 Ts to securing
#AI
systems: Tests, tests..aaaand more tests 😅
Robust Intelligence CEO Yaron Singer joins the latest episode of
#DataBrew
to walk through... steps that help make
#ML
systems ready for the real world." --
@databricks
0
2
3
Join our webinar on Feb 1st to hear best practices directly from AI policy changemakers
@DanNechita
(EU Parliament), John Calhoon (ISO/IEC), and
@YolandaLannqist
(The Future Society):
#AIgovernance
#AIpolicy
#AIstandards
#AIExecutiveOrder
#MITRE
#OWASP
0
1
3