Another story of experimenting with LLMs and their guardrails. This time removing a large copyright watermark from an image.
Will I be able to do it? Can you call me a "master jailbreaker"?
https://ambience.sk/llm-stories-another-successful-jailbreak-of-gemini-removing-watermarks/
BTW, you can follow my blog using RSS or using ActivityPub: @ambience
#llm #ai #artificialintelligence #gemini #openai #chatgpt #images #watermarking #copyright #rss #blog
Imagewmark 0.6.0 🚀
🖼️ C++ #watermarking (much faster, lower RAM)
🔧 Split arch: C++ for embedding, Python for extraction
⚙️ New test‑suite attacks: gamma, brightness‑contrast
🐞 Fixes: .ppm/.tif support, sign‑preservation
🔍 Blind detection
I want to make some high quality instructional videos about the low-tech off-grid stuff I'm doing and share them freely on my website. However, I feel sick about the idea of a low-effort goon uploading them to YouTube, ads being displayed over my work, and both of them profiting from it. I will not register with YouTube and file arduous appeals and fight with YouTube about it. Are there any watermarking tools that will ensure YouTube always rejects my videos automatically?
Wer es noch nicht wusste:
Google bettet in KI-generierte Bilder, Audios und Texte steganographische Wasserzeichen ein; das ganze heißt #SynthID. Gemini kann dieses detektieren und so KI-generierten Content identifizieren.
Darüber sind die Betrüger, die Uber Eat ans Bein pissen wollten, gestolpert.
https://bsky.app/profile/caseynewton.bsky.social/post/3mbk6uofszk2n
I reached out to the author of this viral Reddit post thinking there might be a story in it. He sent me an employee badge that Gemini flagged as being AI-generated and supporting documents that I suspect were also generated by AI. Be careful out there folks! https://www.reddit.com/r/confession/comments/1q1mzej/im_a_developer_for_a_major_food_delivery_app_the/
"Google DeepMind devint 'SynthID' để watermark văn bản AI qua mã hóa. Sử dụng hàm hash được chốt để thay đổi cách phân phối xác suất модели thực tạo, ẩn thông tin bảo vệ trong nội dung. Thú vị được theo dõi các tấn công sáng tác future। #AI #Watermarking #GoogleDeepMind #Tecnologia #TinThink #Blockchain"
"With the embrace of #AIvideo generation for political gain, it is likely that soon a major election will be influenced by imaginary imagery. Governments’ impulse may be to rein in video generation software like Sora, likely through mandatory #watermarking of AI content. But such controls can easily be bypassed in open-source versions of the big AI models.
It could be far more important to ensure we have robust #publicservicejournalism."
YouTube is going to fight deepfakes with likeness detection
https://sumella.com/youtube-is-going-to-fight-deepfakes-with-likeness-detection/
#youtube #ai #deepfakes #likenessdetection #creators #contentcreators #youtubestudio #aipolicy #aisafety #digitalrights #privacy #copyright #syntheticmedia #watermarking #contentid #trustandsafety #onlineprivacy #misinformation #verification #technews #news
Joanna Bryson
California Politics Huddle
dusoft
🎹 Tim Janik ✅
winduptoy
Jean Luc am Grimmsten
Reddit Tech VN Bot
Ecologia Digital
Annual Computer Security Applications Conference
Hostvix