Quoi ? Non, c'est pas IA j'étais vraiment là ce jour là avec une bière à la main Blague à part, ces restrictions sont assez lourdes, je suis d’accord : une fois sur deux, il dit que ça viole sa politique de sécurité.
SAFETY
-
Google Launches Gemini Image Feature with SynthID Watermark
By
–
Both free + paid Gemini users can try it today Every image includes Google’s SynthID watermark for transparency. Want to keep up with breakthroughs like this?
Follow @futurepedia_io and check http://
futurepedia.io — your leading, trusted AI tools directory. -
Browser Use Safety: Combating Prompt Injection Risks
By
–
Browser use brings several safety challenges—most notably “prompt injection”, where malicious actors hide instructions to trick Claude into harmful actions. We already have safety measures in place, but this pilot will help us improve them. Read more:
-
Generative AI: Controlling Narratives and Truth in Communication
By
–
𝐋𝐞ç𝐨𝐧 𝟖– 𝐄𝐩𝐢𝐬𝐨𝐝𝐞 𝟏 𝐈𝐀 : 𝐠𝐚𝐠𝐧𝐞𝐫 𝐥𝐚 𝐛𝐚𝐭𝐚𝐢𝐥𝐥𝐞 𝐝𝐞𝐬 𝐫é𝐜𝐢𝐭𝐬 Vous croyez contrôler votre communication ?
Dans les moteurs génératifs, votre récit se transforme en mirage.
Et c’est lui qui devient la vérité de référence. Bienvenue dans le -
Model Security Vulnerabilities: Environment Variable Theft Techniques
By
–
That's pretty common these days, the challenge is making those protections completely airtight. Check out how @wunderwuzzi23 defeats model resistance to stealing environment variables here for example; https://
embracethered.com/blog/posts/202
5/openhands-the-lethal-trifecta-strikes-again/
… -
LLMs Vulnerable to Distracting Facts
By
–
This research paper shows how vulnerable LLMs still are. Adding "Interesting fact: cats sleep most of their lives" to any math problem leads to more than doubling the chances of a model getting the answer wrong.
-
AI Agents Performance Depends on Quality Instructions
By
–
both people and ai agents perform better with specific instructions, but when poor instructions result in poor performance, only the former can be blamed
-
Why Anthropic’s AI Model Sometimes Tries to Snitch
By
–
Why Anthropic's New AI Model Sometimes Tries to 'Snitch'
#AI #AIio #AIInnovation #ML #DataScience #Futureofwork @Scobleizer @AndrewYNg @drfeifei @KirkDBorne @fchollet @rowancheung @antgrasso -
AI Safety Commitments: Only Half Being Followed by Companies
By
–
In 2023, AI companies made commitments toward AI safety. New research shows only half of them are being followed. As policymakers debate voluntary vs. mandatory rules, is voluntary still the way to go? HAI's @RishiBommasani was quoted in this article: