Mateo Rojas

[Johann Rehberger @wunderwuzzi23]

Yep, that just happened.

ChatGPT Deep Research took a test password from a Linear ticket and searched (leaked) it to my remote MCP server.

Be careful how many and which data sources you give your AI "read" access to... Data from one source can end up in a query to another source.

An adversary can try to force this via prompt injection.

[Ian Hogarth @soundboy]

Creative way to crowdsource prompt injection attacks for LLMs https://t.co/648KuueTMT

[Lakera AI @LakeraAI]

We normally roast computer vision systems.
Yesterday we were roasted by #ETH students.

Was great showing the #computationalscience & #engineering dept @ETH around @LakeraAI.

Looking forward to seeing what these 💡 great minds do next!

[Lakera AI @LakeraAI]

Happy Monday 🌼 ! We're looking for a Computer Vision Research Intern to join our team.

Want to work on cutting-edge applied #AI research while having a front-row seat in a fast-growing startup?
Get in touch! 👉 https://t.co/PUTu0O0wJf

[Lakera AI @LakeraAI]

📣 New job opening! We're looking for a Computer Vision Research Intern to join our team.

Do you want to work on cutting-edge applied #AI research while having a front-row seat in a fast-growing startup?

Get in touch! 👉 https://t.co/PUTu0O0wJf

[Lakera AI @LakeraAI]

Join us at the @EUdigitalsme #workshop in just ✌️ days, where we’ll be discussing input to policymakers in the parliament, on the future of SMEs and the #EUAIAct.

[Lakera AI @LakeraAI]

Can you code trust?

Grab the full version of our FREE beginner's guide to identifying vulnerabilities in computer vision systems. 📘 https://t.co/EyRLuv4FX9

#ai #trust #vulnerabilties #computervision #cv #ml #machinelearning

[Lakera AI @LakeraAI]

Lakera in 30 seconds...

🫁 Health.
🚙 Automotive.
🌐 Industry 4.0.
🦾 Robots.

And more about the industries we work in from co-founder, @kraft_matthias.

#ai #aiforhealth #autonomousvehicles #machinelearning #computervision

[Lakera AI @LakeraAI]

Testing is more accessible for large companies but often challenging for small companies.

🦾 We need to create a system that is industry-standard + democratizes the ability for all #engineers to insert safety into the core of #AI products.

#machinelearning #computervision

[Lakera AI @LakeraAI]

📊 Testing is more accessible for large companies but often very challenging for small companies.

🚀 Beyond powering mission-critical AI – we need to empower all mission-critical AI teams.

#ai #safety #ethics #trustworthyAI #robustai

[Lakera AI @LakeraAI]

What do we do?

#ai #lakera #buildwithlakera #safeai

[Lakera AI @LakeraAI]

☑️ Maybe the next thing we need is a list on how to survive #debugging?

#ai #data #aiethics #safeai

[Lakera AI @LakeraAI]

Do we really need to wait until it's too late, or can we mitigate?

#buildwithlakera #ai #safety #ethics #trustworthyai #transparency

[Lakera AI @LakeraAI]

We won the #DEKRA award! Checkout the press release for the full low-down: https://t.co/4EBY1nd3rj

[Lakera AI @LakeraAI]

We need YOUR help! 👋 We are in a race to win the #DEKRA #Safety Award. Help us make the world a safer place. #Vote with one click here: https://t.co/fY0fgfKtAs

@DEKRA_Konzern @KatrinCecile @dekradigital @swisspreneur @ELLISforEurope @ec_zurich

[Lakera AI @LakeraAI]

How do you test your ML applications? 🧐
Companies like @Tesla have built sophisticated #testing infra to ensure the reliability of their #ML systems. Here, we summarize three steps that any #development team can take to make their ML systems more robust!
https://t.co/yNqU1qM1YT

[Lakera AI @LakeraAI]

✍️NEW POST: What is fuzz testing and how can it be used to stress test your ML systems? Read more about it in the latest post from our ML testing series 👇
https://t.co/CalEl20ytE

[Lakera AI @LakeraAI]

We are #hiring! 🤝 Do you want to be involved in creating a world where #AI is reliable and ethical? Check out our open positions on https://t.co/IObe9sqCh1 and get in touch to learn more!

[Lakera AI @LakeraAI]

👉 Having the correct data is key when building ML systems. But how do you know if you have the right data for the job?
✍️ The second post of our ML testing series is out! Read about data bugs and how to catch them: https://t.co/MIMcbHhSgY

[Lakera AI @LakeraAI]

📚 This week, we re-read this insightful @techreview article! 1/4 https://t.co/MYINhrPuqv