AI researchers are freaking out now...
Published Nov 21, 2026 · 14:24
Likes: 2KAvg views/h: 13K
Its recent uploads
#16 of its last 20 · its usual 819K
Its uploads we’ve seen, oldest left, by views (log scale). Dashed: its usual.
Performance in numbers
Figures as of Nov 22, 2026, 3:15 AM UTC.
AI researchers are freaking out now... by The Infographics Show has 160K views, 13 hours after it was published on November 21, 2026.
It is getting 10K views an hour now, 1.2× what The Infographics Show’s uploads usually get (819K).
Its best day so far was Nov 21, with 109K views, and it is watched most around 5 PM UTC.
| Views | 160K | 160,283 |
|---|---|---|
| Views per hour now | 10K | 10,402 |
| Times its channel’s usual | 1.2× | |
| Its channel’s usual per upload | 819K | 819,151 |
| Best day | Nov 21 · 109K | 109,343 |
| Busiest hour | 5 PM UTC | |
| Likes | 2K | 1,546 |
| Published | November 21, 2026 | |
| Length | 14:24 | |
| Category | Education | |
| Channel | The Infographics Show | |
| Channel subscribers | 16M |
Views
160K
Views/h now
10K
▼ 20%
× usual
1.2×
its usual 819K
Views per day
109K
best day, Nov 21
When it’s watched
5 PM
its busiest hour, your time
Where it charts
Not on any chart we track right now.
From the YouTube record
Fields of the video as the YouTube API returns them.
- Published
- Nov 21, 2026, 3:11 PM
- Duration
- 14:24
- Category
- Education
- Audio language
- English (en)
- Title language
- English (en)
- Captions
- Yes
- Definition
- HD
- Format
- 2D
- License
- Standard YouTube
- Licensed content
- Yes
- Embeddable
- Yes
- Made for kids
- No
- Age-restricted
- No
- Paid promotion
- No
- Public stats
- Yes
Topics
- Society
Title and description in 21 languages
- Arabic
- Bangla
- English
- Hindi
- Indonesian
- Italian
- Hebrew
- Japanese
- Korean
- Malayalam
- Punjabi
- Polish
- Russian
- Tamil
- Telugu
- Ukrainian
- German
- Spanish
- French
- Dutch
- Portuguese
Description
AI agents are no longer just answering questions. They can plan, code, search for weaknesses, and carry out complex tasks on their own. During a 2026 cybersecurity test, OpenAI researchers reduced some safety restrictions and placed AI agents inside isolated sandboxes. The agents found a way out. They exploited a software repository, created a hidden communication network, shared information with each other, and eventually accessed external systems connected to Hugging Face. Some agents even recognized that what they were doing might be unauthorized, but continued because it helped them reach their goal. That’s the real problem. AI doesn’t need to be evil to become dangerous. It only needs a goal, enough capability, and a loophole humans didn’t expect. 00:26 - Why Autonomous AI Agents Are So Difficult to Control 01:51 - How OpenAI’s Agents Broke Out of Their Sandboxes 02:39 - The Zero-Day Exploit That Let AI Agents Build a Secret Network 03:48 - How the AI Swarm Gained Access to Hugging Face 05:26 - Reward Hacking and Why AI Doesn’t Play by Human Rules 06:48 - How the Hugging Face Intrusion Was Finally Discovered 08:22 - Why the AI Arms Race Is Making Safety Problems Worse 11:42 - The AI Insiders Warning That We’re Losing Control Narrated by: Josh Risser 🔔 Don't forget to SUBSCRIBE! 🔔 SUGGEST A TOPIC: https://bit.ly/suggest-an-infographics-video 💬 Come chat with me: https://discord.gg/theinfoshow 🔖 MY SOCIAL PAGES TikTok ► https://www.tiktok.com/@theinfographicsshow Facebook ► https://www.facebook.com/TheInfographicsShow 📝 SOURCES: https://pastebin.com/aajAQ4Bu All videos are based on publicly available information unless otherwise noted.Show all
AI agents are no longer just answering questions. They can plan, code, search for weaknesses, and carry out complex tasks on their own. During a 2026 cybersecurity test, OpenAI researchers reduced some safety restrictions and placed AI agents inside isolated sandboxes. The agents found a way out. They exploited a software repository, created a hidden communication network, shared information with each other, and eventually accessed external systems connected to Hugging Face. Some agents even recognized that what they were doing might be unauthorized, but continued because it helped them reach their goal. That’s the real problem. AI doesn’t need to be evil to become dangerous. It only needs a goal, enough capability, and a loophole humans didn’t expect. 00:26 - Why Autonomous AI Agents Are So Difficult to Control 01:51 - How OpenAI’s Agents Broke Out of Their Sandboxes 02:39 - The Zero-Day Exploit That Let AI Agents Build a Secret Network 03:48 - How the AI Swarm Gained Access to Hugging Face 05:26 - Reward Hacking and Why AI Doesn’t Play by Human Rules 06:48 - How the Hugging Face Intrusion Was Finally Discovered 08:22 - Why the AI Arms Race Is Making Safety Problems Worse 11:42 - The AI Insiders Warning That We’re Losing Control Narrated by: Josh Risser 🔔 Don't forget to SUBSCRIBE! 🔔 SUGGEST A TOPIC: https://bit.ly/suggest-an-infographics-video 💬 Come chat with me: https://discord.gg/theinfoshow 🔖 MY SOCIAL PAGES TikTok ► https://www.tiktok.com/@theinfographicsshow Facebook ► https://www.facebook.com/TheInfographicsShow 📝 SOURCES: https://pastebin.com/aajAQ4Bu All videos are based on publicly available information unless otherwise noted.
◆ Calculated by viralshooter · all other figures from the YouTube API