I'm able to do basic prompt injections with the invisible bytes but I can't get it to work without explicit decoding hints. https://
chatgpt.com/share/67acd3ba
-d234-8007-ad44-ba9d4dfc2920
… The thinking models actually feel a bit more susceptible because they love puzzles and they notice the added bytes and get very
SECURITY
-
Prompt Injection Attempts with Invisible Bytes in AI Models
By
–
-
Unicode Variation Selectors Enable Arbitrary Data Smuggling in Emojis
By
–
UTF-8 I already knew about the "confusables", e.g.: e vs. е. Which look ~same but are different. But you can also smuggle arbitrary byte streams in any character via "variation selectors". So this emoji: 󠅧󠅕󠄐󠅑󠅢󠅕󠄐󠅓󠅟󠅟󠅛󠅕󠅔 is 53 tokens. Yay https://
paulbutler.org/2025/smuggling
-arbitrary-data-through-an-emoji/
… -
Privacy-Protecting Screens Emerge as New Technology Innovation
By
–
This screens will protect your privacy
— Ronald van Loon (@Ronald_vanLoon) 12 février 2025
by @TechBurritoUno#TechForGood #EmergingTech #Innovation #Technology
cc: @paula_piccard @JimMarous @theadamgabriel pic.twitter.com/JDipZkhru0This screens will protect your privacy
by @TechBurritoUno #TechForGood #EmergingTech #Innovation #Technology cc: @paula_piccard @JimMarous @theadamgabriel -
AI Traffic Lights Enforce Motorcycle Safety Helmet Compliance
By
–
Imagine a world where traffic lights ensure your safety.
— Pascal Bornet (@pascal_bornet) 12 février 2025
The traffic light only turns green when motorcyclists wear helmets.
Source: E.Restelica#ai #tech #innovation #security pic.twitter.com/SABCVTRXtqImagine a world where traffic lights ensure your safety. The traffic light only turns green when motorcyclists wear helmets. Source: E.Restelica
#ai #tech #innovation #security -
Senators pressure tech CEOs on illegal content platforms
By
–
Coincidentally, this report comes just days after @MarshaBlackburn and @SenBlumenthal wrote letters to six CEOs expressing concern about anonymous image-sharing websites flagged for hosting CSAM & other problematic content. More info in my Friday story: https://
tinyurl.com/y48m754s -
Data Privacy Security Framework for Critical Industries
By
–
Data privacy and security are paramount across our customers’ industries – including finance, healthcare, and critical infrastructure – so our approach focuses on their unique needs. Access the full framework:
-
Cohere Prioritizes Security and Sustainability in Enterprise AI
By
–
Throughout the summit, Cohere leaders discussed the central roles that security and privacy play in enterprise AI adoption. Cohere also signed the Sustainable AI Coalition along with other industry leaders.
-
Secure AI Frontier Model Framework for Enterprise and Government
By
–
We also released our Secure AI Frontier Model Framework detailing how we develop our AI products with a focus on real-world, practical challenges that enterprises and governments face today:
-
The Future of Security with Drones and Robotics
By
–
The Future of Security with #Drones
— Ronald van Loon (@Ronald_vanLoon) 11 février 2025
via @FrRonconi
#Robotics #AI #MachineLearning #Tech #Technology #Automation #FutureOfWork
cc: @karpathy @chr1sa @johnlegere pic.twitter.com/7Uglc3NOsqThe Future of Security with #Drones
via @FrRonconi #Robotics #AI #MachineLearning #Tech #Technology #Automation #FutureOfWork cc: @karpathy @chr1sa @johnlegere -
Senators Target Adtech Vendors Enabling CSAM Monetization
By
–
ICYMI: Here's my Friday story about @MarshaBlackburn and @SenBlumenthal
's letters to six CEOs — including @doubleverify and @integralads — about concerns about adtech vendors enabling ad monetization on problematic websites including sites flagged for hosting CSAM content.