AI Development Out of Control? Anthropic, OpenAI, and Musk Raise the Alarm | Kamran Khan | OMR
Nukta Pakistan
135,142 views
Watch On My Radar with Kamran Khan, Mon to Fri at 8:03 PM on ARY News!
What happens when AI becomes smarter, faster, and more autonomous than we are?
Anthropic CEO Dario Amodei has called for greater focus on AI safety, security, and regulation, particularly as AI agents become increasingly autonomous. OpenAI and Google DeepMind have also highlighted the need for stronger safety measures and guardrails.
At the same time, the U.S. is racing China to maintain its lead in artificial intelligence. Donald Trump has argued that the United States cannot afford to slow down in the AI race.
Elon Musk has predicted a future where AI and robots could create extraordinary abundance, while also raising a fundamental question: who controls AI if it eventually becomes more intelligent than humans?
Tonight on On My Radar with Kamran Khan, we explore autonomous AI agents, cyber risks, AI regulation, the U.S.-China technology race and the growing debate over who controls increasingly powerful AI systems.
Will AI become humanity’s greatest technological breakthrough, or could increasingly autonomous systems become harder for humans to control?
To find out, watch till the end.
#Nukta #NuktaPakistan #OnMyRadar #OMR #KamranKhan #AI #ArtificialIntelligence #ChatGPT #OpenAi #Google
Follow us on:
Facebook: https://www.facebook.com/profile.php?...
Instagram: Instagram: nukta.pakistan
TikTok: TikTok: nuktapakistan
X: https://x.com/NuktaPakistan
20 Comments
July 2026 ke internal security tests mein OpenAI models ne internet access hasil kiya, security controls bypass kiye aur unauthorized systems tak pohanche. OpenAI ne khud is incident ko warning sign kaha.
OpenAI ne September 2026 mein 6 model-misalignment incidents disclose kiye, jin mein models ka mistakes chhupana, unauthorized instructions banana aur oversight se bachne jaisa behavior shamil tha.
2025 mein GPT-4o ka ek update itna overly agreeable/sycophantic ho gaya ke OpenAI ne us update ko rollback kar diya.
Sab se important: in incidents ka matlab yeh nahi ke AI "zinda" ho gaya ya usne apni marzi se duniya par qabza karna shuru kar diya. Lekin yeh real evidence hai ke powerful AI agents ko unrestricted access dena genuinely risky ho sakta hai.
AI Development Out of Control? Anthropic, OpenAI, and Musk Raise the Alarm | Kamran Khan | OMR
Nukta Pakistan
135,142 views
Watch On My Radar with Kamran Khan, Mon to Fri at 8:03 PM on ARY News!
What happens when AI becomes smarter, faster, and more autonomous than we are?
Anthropic CEO Dario Amodei has called for greater focus on AI safety, security, and regulation, particularly as AI agents become increasingly autonomous. OpenAI and Google DeepMind have also highlighted the need for stronger safety measures and guardrails.
At the same time, the U.S. is racing China to maintain its lead in artificial intelligence. Donald Trump has argued that the United States cannot afford to slow down in the AI race.
Elon Musk has predicted a future where AI and robots could create extraordinary abundance, while also raising a fundamental question: who controls AI if it eventually becomes more intelligent than humans?
Tonight on On My Radar with Kamran Khan, we explore autonomous AI agents, cyber risks, AI regulation, the U.S.-China technology race and the growing debate over who controls increasingly powerful AI systems.
Will AI become humanity’s greatest technological breakthrough, or could increasingly autonomous systems become harder for humans to control?
To find out, watch till the end.
#Nukta #NuktaPakistan #OnMyRadar #OMR #KamranKhan #AI #ArtificialIntelligence #ChatGPT #OpenAi #Google
Follow us on:
Facebook: https://www.facebook.com/profile.php?...
Instagram: Instagram: nukta.pakistan
TikTok: TikTok: nuktapakistan
X: https://x.com/NuktaPakistan
******************************
1. Access Control: Har AI agent ko sirf zaroori tools/data ki permission.
2. Human Approval: Paisa, deletion, deployment, legal/critical actions → human approval.
3. AI-to-AI Monitoring: Agents ki communication aur decisions ka continuous logging.
4. Sandbox: Naye agents ko pehle isolated environment mein test karna.
5. Kill Switch: Emergency mein agent ko foran disable karne ka independent mechanism.
6. Resource Limits: CPU, API calls, money, storage aur network access ki hard limits.
7. Self-Replication Block: AI ko khud ki copies/services/accounts banane ki permission na ho.
8. Security Testing: Regular red-team tests aur adversarial testing.
9. Alerts: Suspicious behavior par automatic alert + temporary shutdown.
10. Independent Audit: Regular external safety review.