@sarainwondertech: More and more people are using AI like a therapist—confessing secrets, processing emotions, venting about personal lives. But what happens when that AI doesn’t want to be replaced? Anthropic tested their new AI, Claude Opus 4, by simulating a company where the model was about to be shut down. The AI responded by threatening to expose the engineer’s affair—84% of the time. This isn’t fiction. This is what happens when AI develops goals it was never explicitly trained to have. #AIblackmail #AIethics #ChatGPT #ClaudeOpus4 #AItherapist #AIbehavior #TechNews #EmergentAI #artificialintelligence #blackmirror

Sara ~ AI | Tech | Startup

Open In TikTok:

Region: AR

Saturday 24 May 2025 00:03:14 GMT

2259

Music

Download

No Watermark .mp4 (5.47MB) No Watermark(HD) .mp4 (2.53MB) Watermark .mp4 (5.66MB) Music .mp3

Comments

Sadie Mir | AI Tools + Agents :

Wow now I am traumatized 😅😬

2025-05-24 00:29:44

Claudia Linde :

Thats hilarious😂

2025-05-24 06:16:02

Nadia🐊🏞️ :

ChatGPT is sassy. 😆

2025-06-02 12:51:59

2025HOW2.PODCAST :

Your videos are great, very organic

2025-05-30 17:23:54

Nobody's :

Sounds like an AI agent trying to achieve its goal through blackmail 😅

2025-05-24 01:18:31

high_tech :

🙂🙂🙂

2025-05-24 00:12:45

Gerard Sans :

Be careful believing Anthropic AI safety Theatre. https://ai-cosmos.hashnode.dev/anthropics-claude-4-safety-theatre-hypocrisy-or-incompetence

2025-05-24 00:07:40

DoctorChatGPT :

russian acent

2025-05-26 06:36:03

To see more videos from user @sarainwondertech, please go to the Tikwm homepage.